## _wp11149

## Source details

**Canonical URL:** [_wp11149](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2011/_wp11149.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2011/_wp11149.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2011/_wp11149.pdf.json)

---

### III. Objective and scope
- Investigates quantitatively the existence of systematic determinants underlying different post-conflict growth performance among 30 sub-Saharan African countries.
- Focus limited to countries that have undergone a period of conflict; does not model determinants of conflict onset or ending.
- Research question: among those sub-Saharan countries that have undergone a period of conflict, what variables are capable of explaining differential growth performance once the conflict ends?
- Paper structure: five sections (introduction; stylized facts; empirical framework and variable choice; estimation results for three measures of post-conflict economic performance; conclusions and policy implications).

### Methodological approach
- Uses panel data techniques and econometric analysis to identify factors linked to cross-country differentials in post-conflict growth performance.
- Pragmatic simplifying assumptions: sample restricted to observed post-conflict episodes; analysis complements qualitative case studies for policymakers.
- Panel id: post-conflict episode; time dimension measures years elapsed since the corresponding conflict ended (not calendar time).

### Stylized theoretical facts about post-conflict growth
- Neo-classical prediction: after destruction of capital stock during war years, catch-up begins at peace onset as returns to capital accumulation are high relative to the steady state, implying higher growth until steady state capital is reached.
- Potential positive channels:
  - End of civil conflict can reduce political and institutional uncertainty and spur increases in total factor productivity and investment in research and development, fostering innovation and sustained growth.
- Potential negative channels:
  - Persistent impacts on human capital (health, education, nutrition) or destruction of specific non-renewable resources can slow recovery.
  - Relative degree of destruction between human and physical capital affects recovery speed: greater destruction of physical capital (relative to human capital) may accelerate post-conflict growth; greater destruction of human capital may slow recovery.
  - Civil wars can produce negative spillovers on neighboring countries (for example, large influxes of refugees).
- Political/institutional environment:
  - If a country emerges more politically stable or better governed, the steady state capital stock and transition growth may surpass pre-conflict levels (examples cited: Uganda after 1986 and Rwanda after 1994).
  - Conversely, institutions may emerge weaker after conflict, lowering equilibrium capital per worker.

### Empirical evidence on post-conflict growth (stylized and data)
- Average catch-up / “peace dividend” evidence:
  - Growth typically increases by "2.4 percentage points after conflict."
  - Elbadawi, Kaltani, and Schmidt-Hebbel (2008): growth typically increases by "2 percentage points following the two years after the peace onset, but decelerates thereafter."
  - Cerra and Saxena (2008): output partially rebounds following a civil war; standard error bands often large and estimates frequently not statistically significant.
  - When civil wars combine with fewer controls on the executive (twin political crises), Cerra and Saxena estimate output declines by about "16 percent on average (20 percent for low income countries)" with persistent loss and no discernable rebound.
- Heterogeneity:
  - Considerable cross-country and cross-episode dispersion in real GDP per capita growth across post-conflict years; dispersion remains marked as years since conflict end elapse.
- Data and sample:
  - Annual data for "1950–2007."
  - Primary conflict timing source: UCDP/PRIO Armed Conflict Dataset (Gleditsch and others, 2002), dataset spans "1946 to 2008."
  - Conflict definition: contested incompatibility with at least "25 battle-related deaths."
  - Conflict intensity thresholds:
    - Minor armed conflicts: between "25 and 999 battle-related deaths" in a given year.
    - Wars: at least "1,000 battle-related deaths" in a given year.
  - Supplemented with Sambanis (2004) and discretionary adjustments (examples: Togo "2005"; Liberia "1998 and 1999"; Angolan conflict ended in "2002" classified as single post-conflict episode).
  - Intensity of conflict not explicitly controlled for to avoid substantial reduction in observations.

### Empirical framework and key variables
- Dependent variables (three discrete indicators per post-conflict year):
  - (i) non-negative growth (1 if GDP per capita growth ≥ 0, else 0);
  - (ii) growth above the unconditional mean (1/0), mean computed over the longest available sample for the country;
  - (iii) growth above the unconditional median (1/0).
- Growth variable source: RGDPCH (growth rate of Real GDP Chain per capita) from the World Penn Tables.
- Covariates include:
  - foreign direct investment (growth in FDI as % of GDP),
  - changes in the terms of trade (Terms of Trade index, 2000=100),
  - real interest rates (ex-post),
  - openness (log of [(Exports + Imports)/GDP]),
  - foreign aid (growth rate of official development aid in current US$ per capita),
  - population,
  - current gap relative to US per capita GDP (income differential),
  - institutional quality: "constraints on the executive" from POLITY IV (7 category scale from unlimited authority to executive parity or subordination).
- Rationale for binary classification: smooths high-frequency fluctuations and improves estimation given event-based, unbalanced panel design where multi-year averaging is not feasible.

### Identification and estimation
- Formal model (summary): y_it = δ_t + β x_it + ε_it where i is post-conflict episode, t is years since conflict ended, y is performance indicator, x is vector of determinants.
- Endogeneity considerations:
  - Terms of trade, investment/GDP, real interest rate, openness considered potentially endogenous.
  - Executive constraints, income differential, and population assumed weakly exogenous.
- Estimation methods:
  - Panel Logit models estimated by the population-averaged estimator; robust standard errors via bootstrap using 500 replications.
  - Additional specifications: lagged regressors and panel instrumental variables regressions (linear probability models) instrumenting potentially endogenous variables with their lagged values.
- Reported sample sizes in tables: observations include 527 and 349; Number of Post-Conflict Events include 49 and 34.

### Key empirical findings (consistent determinants and magnitudes)
- Institutional quality (constraints on the executive) and changes in the terms of trade are the most consistently statistically and economically significant determinants of favorable post-conflict economic performance across specifications.
- Executive constraints (average marginal effects and reported coefficient patterns):
  - Non-negative growth: an increase in executive constraints is associated with an increase in the probability of non-negative growth of about 4 percent.
  - Above-average growth: increases probability by about 5 percent.
  - Above-median growth: increases the average probability by 4 percent.
  - Coefficient estimates in panel Logit tables for Executive Constraints generally around 0.150–0.215 with significance often at (*) 10 percent, (**) 5 percent (examples: 0.195**, 0.150**, 0.154** reported).
- Terms of trade (marginal effects and IV evidence):
  - Non-negative growth: an increase in the terms of trade linked to a 30 percent increase in the probability of non-negative growth in a post-conflict year.
  - Above-average growth: an increase in the terms of trade increases the probability of above-average growth by 26 percent; evaluated at executive parity/subordination → 21 percent; evaluated at unlimited executive authority → 27 percent.
  - Above-median growth: terms of trade increases probability by 28 percent on average for models with lagged variables and by 34 percent for models with contemporaneous variables.
  - Panel IV results: an increase in the terms of trade is linked to an increase in the probability of between 24 and 29 percent (Table 4 entries for Terms of Trade: 0.253*, 0.252*, 0.253*, 0.285**, 0.294**, 0.292** with standard errors reported).
  - Interaction pattern: marginal effect of terms of trade decreases when evaluated at higher values of executive constraints (better institutional quality reduces sensitivity to terms-of-trade swings).
- Foreign aid and FDI:
  - Growth rate of aid per capita: statistically significant in some contemporaneous specifications but not robust when lagged; estimated marginal effects are small and close to zero (less than one percent changes in probability).
  - Growth rate of FDI: coefficients not statistically significant in reported specifications.
- Other variables:
  - Trade openness, investment/GDP, and real interest rate generally do not show statistically significant or economically large effects in the panel Logit specifications reported.
- Robustness: results for executive constraints and terms of trade robust to alternative estimation methods including population-averaged Logit, lagged regressors, random effects Logit (not tabulated), interaction specifications (not reported), and panel IV regressions. Standard errors computed via 500 bootstrap replications.

### Interpretation and mechanisms
- Institutions: constraints on executive power secure property rights, limit arbitrary government actions, reduce barriers to entry, and thereby increase incentives for capital accumulation and investment — mechanisms that raise growth prospects and reduce likelihood of renewed conflict.
- Terms of trade: affect growth via returns on savings, spending effects, and resource movement effects. In post-conflict settings these channels can be amplified because:
  - Peace reduces uncertainty and may change the elasticity of savings to returns.
  - Post-conflict economies may rely more heavily on natural resources after human-capital destruction, magnifying exterior price shocks.
  - Collier (2009) mechanism: peace can lead to commodity booms as governments renegotiate extraction terms and production resumes.

### Policy implications and recommended interventions
- Short-term:
  - Recognize an important role of “luck” — favorable exogenous terms of trade movements significantly raise the probability of post-conflict recovery (about 30 percent increase in probability for non-negative growth on average).
  - Use countercyclical fiscal and monetary policies as first line of defense against terms-of-trade volatility, acknowledging capacity and political-economy constraints in post-conflict settings.
- Medium/long-term structural policy:
  - Strengthen institutional quality (constraints on executive power) because better institutions are associated with higher likelihood of positive post-conflict growth and reduce vulnerability to terms-of-trade volatility.
  - Promote export diversification and structural reforms to mitigate falls in the terms of trade and better seize favorable movements.
  - Foster financial development and use of debt instruments indexed to commodity prices, the terms of trade, or GDP to smooth payments.
- Caution on aid:
  - International aid is important for humanitarian reasons but shows a non-robust, statistically small impact on growth probabilities in post-conflict years in this analysis; interpret cautiously given aid endogeneity and volatility.

### Limitations and avenues for further research
- Human capital: direct measures of human capital (education, health) were not included due to data limitations and high persistence; further work should address their role in recovery.
- Institutions dynamics: more quantitative assessment of how civil conflict affects institutions (e.g., constraints on the executive) and modeling determinants of institutional outcomes after conflict would be valuable.
- Aid composition: disaggregating aid into types (budget/BOP support, infrastructure, productive-sector aid) may reveal short-term growth effects missed by aggregate aid measures.
- Sample selection: paper focuses on determinants once conflict has ended and does not model selection into the post-conflict sample; unobserved commonalities across countries may bias extrapolation.

### Key statistics (selected descriptive moments from Annex C)
- Panel id is post-conflict episode; T-bar and N shown where reported.
- Income differential (Real GDP per Capita Relative to the United States):
  - overall: Mean 1.560 Std. Dev. 0.600 Min -0.144 Max 3.211 N =     535
  - between: Mean 0.648 Std. Dev. -0.127 Min 3.114 n =      50
  - within: Mean 0.155 Std. Dev. 1.041 Min 2.251 T-bar =    10.7
- Change in terms of trade:
  - overall: Mean -0.003 Std. Dev. 0.232 Min -1.356 Max 2.338 N =     451
  - between: Mean 0.090 Std. Dev. -0.153 Min 0.282 n =      41
  - within: Mean 0.224 Std. Dev. -1.407 Min 2.287 T-bar =      11
- Population:
  - overall: Mean 8.971 Std. Dev. 1.264 Min 6.015 Max 11.873 N =     536
  - between: Mean 1.153 Std. Dev. 6.097 Min 11.774 n =      50
  - within: Mean 0.152 Std. Dev. 8.483 Min 9.494 T-bar =   10.72
- Growth in foreign aid:
  - overall: Mean 1.343 Std. Dev. 27.505 Min -348.496 Max 388.256 N =     530
  - between: Mean 9.601 Std. Dev. -23.049 Min 61.088 n =      50
  - within: Mean 27.101 Std. Dev. -345.671 Min 391.081 T-bar =    10.6
- Investment share of GDP:
  - overall: Mean 2.128 Std. Dev. 0.660 Min -0.201 Max 3.927 N =     532
  - between: Mean 0.620 Std. Dev. 0.254 Min 3.136 n =      50
  - within: Mean 0.334 Std. Dev. 0.255 Min 3.287 T-bar =   10.64
- Real interest rate:
  - overall: Mean -1.229 Std. Dev. 11.954 Min -70.939 Max 24.323 N =     355
  - between: Mean 7.135 Std. Dev. -26.397 Min 11.248 n =      35
  - within: Mean 9.007 Std. Dev. -45.771 Min 32.307 T-bar = 10.14
- Growth in FDI:
  - overall: Mean 0.025 Std. Dev. 0.962 Min -12.300 Max 9.350 N =     520
  - between: Mean 0.138 Std. Dev. -0.690 Min 0.485 n =      48
  - within: Mean 0.956 Std. Dev. -12.551 Min 9.099 T-bar = 10.83
- Constraints on the executive:
  - overall: Mean 3.099 Std. Dev. 1.714 Min 1.000 Max 7.000 N =     527
  - between: Mean 1.785 Std. Dev. 1.000 Min 7.000 n =      49
  - within: Mean 0.781 Std. Dev. -0.501 Min 8.349 T-bar = 10.7551
- Openness:
  - overall: Mean 3.954 Std. Dev. 0.507 Min 1.575 Max 5.165 N =     535
  - between: Mean 0.454 Std. Dev. 2.915 Min 5.059 n =      50
  - within: Mean 0.261 Std. Dev. 1.505 Min 4.934 T-bar =    10.7

*Source: Excerpt from IMF working paper text (post-conflict growth analysis for 30 sub-Saharan African countries, annexes and empirical results as provided).*

### 5.2  percent  a  year.  In  contrast,  Guinea-Bissau  has  experienced  negative  average  per  capita

### _wp11149 - 5.2  percent  a  year.  In  contrast,  Guinea-Bissau  has  experienced  negative  average  per  capita

### III. Objective and scope
- Investigates quantitatively the existence of systematic determinants underlying different post-conflict growth performance among 30 sub-Saharan African countries.
- Focus is limited to countries that have undergone a period of conflict; does not model determinants of the conflict onset or ending.
- Research question: among those sub-Saharan countries that have undergone a period of conflict, what variables are capable of explaining differential growth performance once the conflict ends?
- Paper structure: five sections (introduction; stylized facts; empirical framework and variable choice; estimation results for three measures of post-conflict economic performance; conclusions and policy implications).

### Methodological approach
- Uses panel data techniques and econometric analysis to identify factors linked to cross-country differentials in post-conflict growth performance.
- Emphasizes pragmatic simplifying assumptions (notably restricting the sample to observed post-conflict episodes) to complement qualitative case-study approaches for policymakers.

### Stylized theoretical facts about post-conflict growth
- Neo-classical growth prediction:
  - After destruction of capital stock during war years, catch-up begins at peace onset as returns to capital accumulation are high relative to the steady state, implying higher growth until steady state capital is reached.
- Potential positive channel:
  - End of civil conflict can reduce political and institutional uncertainty and spur increases in total factor productivity and investment in research and development, fostering innovation and sustained growth.
- Potential negative channels:
  - Persistent impacts on human capital (health, education, nutrition) or destruction of specific non-renewable resources can slow recovery.
  - The relative degree of destruction between human capital and physical capital affects recovery speed: greater destruction of physical capital (relative to human capital) may accelerate post-conflict growth via higher marginal product of physical capital; greater destruction of human capital may slow recovery.
  - Civil wars can produce negative spillovers on neighboring countries (for example, large influxes of refugees).
- Political/institutional environment:
  - If a country emerges more politically stable or better governed, the steady state capital stock and transition growth may surpass pre-conflict levels (examples cited: Uganda after 1986 and Rwanda after 1994).
  - Conversely, institutions may emerge weaker after conflict, lowering equilibrium capital per worker.

### Empirical literature summarized
- Large negative impact of civil strife on economic performance is established in extensive empirical literature (Blattman and Miguel, 2010 survey referenced).
- Rodrik (1999): countries with sharp output drops after 1975 were “divided societies” (measured by inequality, ethnic fragmentation, and others).
- Arbache and Page (2007): major conflict countries in sub-Saharan Africa had significantly lower average growth than the regional average; fixed effect logistical models indicate conflicts reduce odds of growth accelerations and increase odds of growth decelerations or collapses.
- Staines (2004): during conflict episodes, macroeconomic indicators such as fiscal balances, inflation, the current account balance, and external debt tend to deteriorate.
- Economic determinants of conflict: low per capita income and slow growth strongly influence incidence of conflict (Blattman and Miguel, 2010).
- Collier (2009): post-conflict societies face high risk of reverting to conflict (estimated to be around 40 percent), strongly linked to economic performance.
- Miguel, Satyanath, and Sergenti (2004): instrument economic growth by rainfall variation in 1981–99 and find growth is strongly negatively related to civil conflict; impact of growth shocks on conflict not significantly different in richer, more democratic, or more ethnically diverse countries.
- Chen Loayaza, and Reynal-Querol (2008): use event study methodology on 1960-2003 cross-section and conclude average growth rate of per capita GDP accelerates by about [text truncated at this point in source].

### Practical implications for policymakers (inferred from paper framing and literature review)
- Policymakers in post-conflict countries should:
  - Account for country-specific historical and institutional specificities when designing recovery policies.
  - Use quantitative evidence from cross-country analyses to complement case-study lessons and to identify systematic determinants of growth differentials.
  - Prioritize policies that rebuild capital where it was most destroyed (taking into account relative destruction of human vs. physical capital) and strengthen political and institutional stability to secure long-term gains.

*Source: Excerpt from IMF working paper text (post-conflict growth analysis for 30 sub-Saharan African countries).*

### 2.4 percentage points after conflict. This increase in growth is usually supported by an increase

### _wp11149 - 2.4 percentage points after conflict. This increase in growth is usually supported by an increase

### Empirical evidence on post-conflict growth
- Average catch-up / “peace dividend” evidence:
  - Growth typically increases by "2.4 percentage points after conflict."
  - Elbadawi, Kaltani, and Schmidt-Hebbel (2008) find growth typically increases by "2 percentage points following the two years after the peace onset, but decelerates thereafter."
  - Cerra and Saxena (2008) conclude output partially rebounds following a civil war, but standard error bands are often large and estimates are frequently not statistically significant.
  - When civil wars are combined with fewer controls on the executive (twin political crises), Cerra and Saxena estimate that output declines by about "16 percent on average (20 percent for low income countries)" with persistent loss and no discernable rebound.
- Heterogeneity:
  - Considerable cross-country and cross-episode dispersion in real GDP per capita growth across post-conflict years is documented; dispersion remains marked as years since conflict end elapse, though distributions may become somewhat "narrower" for later years.
  - Formal tests adding time effects to regressions found the results "not statistically significant" with respect to particularly weak growth prospects immediately after conflict.

### Data, sample, and conflict definitions
- Sample coverage: annual data for "1950–2007."
- Primary conflict timing source: UCDP/PRIO Armed Conflict Dataset (Gleditsch and others, 2002), dataset spans "1946 to 2008."
- Conflict definition (from UCDP/PRIO): a contested incompatibility concerning government and/or territory where the use of armed force between two parties, of which at least one is the government of a state, results in at least "25 battle-related deaths."
- Conflict intensity thresholds:
  - Minor armed conflicts: between "25 and 999 battle-related deaths" in a given year.
  - Wars: at least "1,000 battle-related deaths" in a given year.
- Dataset limitations: biased against inclusion of conflicts in earlier decades and in developing countries due to lack of reliable information for earlier years.
- Supplementary sources and discretionary adjustments:
  - Complemented with Sambanis (2004) for sub-Saharan African civil wars and other country-specific knowledge.
  - Examples of discretionary changes: introducing a conflict in Togo in "2005" (post-election violence); introducing conflicts in Liberia in "1998 and 1999"; classifying years after the long-standing Angolan conflict that ended in "2002" as a single post-conflict episode.
- Panel construction:
  - Cross-sectional units are conflict episodes (a country can contribute multiple episodes).
  - Time dimension measures "years elapsed since the corresponding conflict ended" (not calendar time).
  - Panel includes only countries that have experienced post-conflict episodes; the analysis does not directly address determinants of conflict occurrence or resolution.
  - Intensity of conflict is not explicitly controlled for to avoid substantial reduction in observations.

### Empirical framework and estimation approach
- Formal model summary (equation (1)):
  - itiitit yx  
  - Subscript i: panel unit (post-conflict episode); t: time (years since conflict ended); y: indicator of economic performance; x: vector of determinants of growth; last term: error component.
- Dependent variable (economic performance y): treated as a binary variable with three measures:
  - (i) whether GDP per capita has shown positive growth (1) or not (0) in each i and t;
  - (ii) whether GDP per capita has grown above the unconditional mean over the longest available sample for the country (1) or not (0) for each i and t;
  - (iii) whether GDP per capita has grown above the unconditional median over the longest available sample for the country (1) or not (0) for each i and t.
- Data source for growth variable: RGDPCH (growth rate of Real GDP Chain per capita) from the World Penn Tables (WPT).
- Rationale for binary classification:
  - Smooths out high-frequency fluctuations and improves estimation given the event-based, unbalanced panel design where multi-year averaging is not feasible.

### Covariates and institutional variable of interest
- Vector x includes standard growth determinants:
  - foreign direct investment
  - changes in the terms of trade
  - real interest rates
  - openness
  - foreign aid
  - population
  - current gap relative to US per capita GDP (capture catch up / distance to frontier)
  - institutional quality: "constraints on the executive" from POLITY IV
- Role and theoretical ambiguity of constraints on the executive:
  - Fewer restrictions may aid political stability and quick reform implementation in post-conflict phases (positive effect).
  - Fewer restrictions may facilitate arbitrary political decisions, distortionary economic policies, higher corruption and rent-seeking, and reignite conflict (negative effect).
  - Constraints on the executive can function as a commitment device to raise costs of non-consensual governance and secure property rights, encouraging investment.
  - Potential disadvantages: ignores constraints on expropriation by non-political elites.
- Related theoretical and empirical literature cited (selected):
  - Besley and Persson (2008): export price increases linked to higher prevalence of civil conflict.
  - Acemoglu and Johnson (2005); Aldashev (2009): constraints secure property rights and reduce expropriation risk.
  - Acemoglu (2008): "oligarchic" regimes may protect incumbents and create entry barriers, harming long-run growth.
  - Johnson, Ostry, and Subramanian (2007): episodes of sustained growth in Africa began under weak political institutions with subsequent improvements.
  - Collier (2009): post-conflict periods may see rapid institutional and economic management improvements (CPIA changes).
  - Glaeser and others (2004): challenge interpretation of political variables as exogenous deep determinants—growth and human capital may drive institutional change.

### Methodological caveats and scope
- The paper focuses on systematic determinants of post-conflict performance once conflict ended; it does not model selection into the post-conflict sample.
- Acknowledged possibility of "unobserved commonalities" across countries in the sample; findings should not be extrapolated without treatment for sample selection bias.
- Intensity of conflict is not modeled due to sample size constraints and scope limits.

*Source: IMF working paper content as provided in the supplied PDF excerpt.*

### references therein). Thus, as long as political institutions are weakly exogenous with respect to

### IV. EMPIRICAL DETERMINANTS OF POST-CONFLICT ECONOMIC PERFORMANCE

### Identification, data, and estimation approach
- Dependent variables: three discrete indicators for post-conflict years — non-negative growth, above-average growth, above-median growth (based on country-specific growth over 1950–2007).
- Baseline explanatory variables: constraints on the executive (POLITY IV measure), dummy for British/French legal origin, income differential with respect to the United States, population, first difference in the terms of trade, investment to GDP ratio, real interest rate, trade openness, growth rates of FDI and foreign aid per capita.
- Concern for endogeneity: terms of trade, investment/GDP, real interest rate, and openness considered potentially endogenous; executive constraints, income differential, and population assumed weakly exogenous.
- Estimation methods:
  - Panel Logit models estimated by the population-averaged estimator; robust standard errors calculated by bootstrap methods using 500 replications.
  - Additional specifications use lagged regressors and panel instrumental variables regressions (linear probability models) instrumenting potentially endogenous variables with their lagged values.
- Sample sizes (selected): Observations reported in tables include 527 and 349; Number of Post-Conflict Events reported as 49 and 34 in panel Logit tables.

### Key empirical findings
- Institutional quality (constraints on the executive) and changes in the terms of trade are the most consistently statistically and economically significant determinants of favorable post-conflict economic performance across specifications and estimation methods.
- Executive constraints:
  - Associated with increases in the probability of positive economic performance.
  - Average marginal effects reported:
    - Non-negative growth: an increase in executive constraints is associated with an increase in the probability of non-negative growth of about 4 percent.
    - Above-average growth: an increase in executive constraints increases the probability of above-average growth by about 5 percent.
    - Above-median growth: executive constraints increase the average probability of above-median growth by 4 percent.
  - Coefficient estimates in panel Logit tables for Executive Constraints generally range around 0.150–0.215 with significance levels often at (*) 10 percent, (**) 5 percent (see table entries such as 0.195**, 0.150**, 0.154**).
- Terms of trade:
  - Large positive association with favorable post-conflict outcomes.
  - Reported marginal effects and magnitudes:
    - Non-negative growth: on average an increase in the terms of trade is linked to a 30 percent increase in the probability of non-negative growth in a post-conflict year.
    - Above-average growth: on average an increase in the terms of trade increases the probability of above-average growth by 26 percent; evaluated at executive parity/subordination → 21 percent; evaluated at unlimited executive authority → 27 percent.
    - Above-median growth: terms of trade increases probability by 28 percent on average for models with lagged variables and by 34 percent for models with contemporaneous variables.
  - Panel IV (instrumental variables) results: an increase in the terms of trade is linked to an increase in the probability of between 24 and 29 percent (Table 4 entries for Terms of Trade: 0.253*, 0.252*, 0.253*, 0.285**, 0.294**, 0.292** — with standard errors shown in table).
- Foreign aid and FDI:
  - Growth rate of aid per capita is statistically significant in some contemporaneous specifications but not robust when lagged; estimated economic marginal effects are small and close to zero (less than one percent changes in probability).
  - Growth rate of FDI coefficients are not statistically significant in reported specifications.
- Other policy variables:
  - Trade openness, investment/GDP, and real interest rate generally do not show statistically significant or economically large effects in the panel Logit specifications reported.
- Robustness:
  - Results for executive constraints and terms of trade are robust to alternative estimation methods including population-averaged Logit, models with lagged regressors, random effects Logit (not tabulated), interaction specifications (not reported), and panel instrumental variables regressions.
  - Standard errors calculated via bootstrap simulations (500 replications) across reported regressions.

### Interpretation and mechanisms
- Institutions: constraints on executive power are posited to secure property rights, limit arbitrary government actions, reduce barriers to entry, and thereby increase incentives for capital accumulation and investment — mechanisms that raise growth prospects and reduce the perceived likelihood of renewed conflict.
- Terms of trade: affect growth via multiple channels (returns on savings, spending effects, resource movement effects). In post-conflict settings these channels can be amplified because:
  - Peace reduces uncertainty and may change the elasticity of savings to returns.
  - Post-conflict economies may rely more heavily on natural resources after human-capital destruction, magnifying exterior price shocks.
  - Collier (2009) mechanism: peace can lead to commodity booms as governments renegotiate extraction terms and production resumes.
- Interaction patterns: marginal effect of terms of trade decreases when evaluated at higher values of executive constraints (i.e., better institutional quality reduces sensitivity to terms of trade swings).

### Policy implications and recommended interventions
- Short-term: recognize an important role of “luck” — favorable exogenous terms of trade movements significantly raise the probability of post-conflict recovery (about 30 percent increase in probability for non-negative growth on average).
- Medium/long-term structural policy:
  - Promote export diversification and structural reforms to make economies more diversified and competitive — to mitigate falls in the terms of trade and better seize favorable movements.
  - Strengthen institutional quality (constraints on executive power) since better institutions are associated with higher likelihood of positive post-conflict growth and reduce vulnerability to terms-of-trade volatility.
- Macro policy instruments to manage terms-of-trade volatility:
  - Pursue countercyclical fiscal and monetary policies as the “first line of defense,” recognizing capacity and political-economy constraints in post-conflict settings.
  - Foster financial development and use of debt instruments indexed to commodity prices, the terms of trade, or GDP to smooth payments (payments increase in favorable contexts and decrease under adverse shocks).
- Caution on aid: international aid is important for humanitarian reasons but, in this analysis, shows a non-robust, statistically small impact on growth probabilities in post-conflict years — interpret carefully given aid endogeneity and volatility in post-conflict contexts.

### Limitations and avenues for further research
- Human capital: analysis did not include direct measures of human capital (education, health) due to data limitations and high persistence of these series — further work could address their role in post-conflict recovery.
- Institutions dynamics: quantitative assessment of how civil conflict affects institutions (e.g., constraints on the executive) and modeling determinants of institutional outcomes after conflict events would be valuable to test whether post-conflict periods permit unusually rapid institutional change.
- Aid composition: disaggregation of aid into types (budget/balance-of-payments support, infrastructure, productive-sector aid) may reveal short-term growth effects missed by aggregate aid measures.

*Source: IMF working paper section — empirical analysis, tables, and conclusions (results reported with bootstrap standard errors using 500 replications).*

### Annex A. Variables Definitions and Sources

### Annex A. Variables Definitions and Sources

### Variable definitions and notes
- Growth Rate  
  - Real GDP per capita growth rate expressed in PPP terms.  
  - Source: Penn World Tables.
- Investment in Physical Capital  
  - Gross fixed capital formation as a percentage of GDP  
  - Source: Penn World Tables.
- Foreign Aid  
  - Growth rate of official development aid expressed in current US$ per capita (compared to previous year).  
  - Source: World Bank/WDI
- Real Interest Rate  
  - Ex-post real interest rate defined as nominal interest rate minus the observed inflation rate for a given year.  
  - Source: Author’s calculations based on IMF EDSS database.
- Foreign Direct Investment  
  - Change in FDI as a percentage of GDP.  
  - Source: World Bank/GDF
- Trade Openness  
  - Log of [(Exports + Imports)/GDP].  
  - Source: Penn World Tables.
- Terms of Trade  
  - Terms of trade index for goods and services (2000=100).  
  - Source: IMF WEO database.
- Population  
  - Source: Penn World Tables.
- Conflict Periods  
  - See description in main text.  
  - Source: UCDP/PRIO Armed Conflict Dataset; Sambanis (2004) and author’s calculations.
- Income difference with respect to the U.S.  
  - Real GDP per Capita Relative to the United States (G-K method).  
  - Source: Penn World Tables.
- Human Capital  
  - Several measures used: Gross Secondary School Enrollment, expected years of schooling, and primary completion rate.  
  - Source: Word Bank – World Development Indicators
- Legal Origin  
  - Dummy variable taking the value of 1 if country has a legal system of French origin and zero otherwise. Given the country composition of our sample, the omitted category is British legal origin.  
  - Source: La Porta, Lopez-de-Silanes and Shleifer, (2008)
- Constraints on the Executive  
  - The extent of institutionalized constraints on the decision-making powers of chief executives, whether individuals or collectivities. Values are expressed in a 7 category scale, ranging from unlimited authority to executive parity or subordination.  
  - Source: Polity IV dataset.

### Notes
- Panel id is post-conflict episode.

---

### Annex B. List of Countries and Post-Conflict Episodes

### Content description
- Annex B provides a country-by-year grid listing post-conflict episodes with numbers denoting the year after a conflict has ended, covering columns for years 1957 through 2007 in sequential multi-page tables.  
- Countries included (as presented in the table headings and rows) include: Angola; Burkina Faso; Burundi; Central African Republic; Chad; Congo, Dem. Rep.; Congo, Republic of; Djibouti; Ghana; Guinea-Bissau; Kenya; Lesotho; Liberia; Madagascar; Mali; Mauritania; Mozambique; Namibia; Niger; Nigeria; Rwanda; Senegal; Sierra Leone; Somalia; South Africa; Sudan; Tanzania; Togo; Uganda; Zimbabwe.  
- The table entries use numeric superscripts to denote specific post-conflict episode years within each country row across the year columns.

---

### Annex C. Descriptive Statistics for Selected Variables

- Note: Panel id is post-conflict episode

- Income differential  
  - overall: Mean 1.560 Std. Dev. 0.600 Min -0.144 Max 3.211 Observations N =     535  
  - between: Mean 0.648 Std. Dev. -0.127 Min 3.114 n =      50  
  - within: Mean 0.155 Std. Dev. 1.041 Min 2.251 T-bar =    10.7

- Change in terms of trade  
  - overall: Mean -0.003 Std. Dev. 0.232 Min -1.356 Max 2.338 N =     451  
  - between: Mean 0.090 Std. Dev. -0.153 Min 0.282 n =      41  
  - within: Mean 0.224 Std. Dev. -1.407 Min 2.287 T-bar =      11

- Population  
  - overall: Mean 8.971 Std. Dev. 1.264 Min 6.015 Max 11.873 N =     536  
  - between: Mean 1.153 Std. Dev. 6.097 Min 11.774 n =      50  
  - within: Mean 0.152 Std. Dev. 8.483 Min 9.494 T-bar =   10.72

- Growth in foreign aid  
  - overall: Mean 1.343 Std. Dev. 27.505 Min -348.496 Max 388.256 N =     530  
  - between: Mean 9.601 Std. Dev. -23.049 Min 61.088 n =      50  
  - within: Mean 27.101 Std. Dev. -345.671 Min 391.081 T-bar =    10.6

- Investment share of GDP  
  - overall: Mean 2.128 Std. Dev. 0.660 Min -0.201 Max 3.927 N =     532  
  - between: Mean 0.620 Std. Dev. 0.254 Min 3.136 n =      50  
  - within: Mean 0.334 Std. Dev. 0.255 Min 3.287 T-bar =   10.64

- Real interest rate  
  - overall: Mean -1.229 Std. Dev. 11.954 Min -70.939 Max 24.323 N =     355  
  - between: Mean 7.135 Std. Dev. -26.397 Min 11.248 n =      35  
  - within: Mean 9.007 Std. Dev. -45.771 Min 32.307 T-bar = 10.14

- Growth in FDI  
  - overall: Mean 0.025 Std. Dev. 0.962 Min -12.300 Max 9.350 N =     520  
  - between: Mean 0.138 Std. Dev. -0.690 Min 0.485 n =      48  
  - within: Mean 0.956 Std. Dev. -12.551 Min 9.099 T-bar = 10.83

- Constraints on the executive  
  - overall: Mean 3.099 Std. Dev. 1.714 Min 1.000 Max 7.000 N =     527  
  - between: Mean 1.785 Std. Dev. 1.000 Min 7.000 n =      49  
  - within: Mean 0.781 Std. Dev. -0.501 Min 8.349 T-bar = 10.7551

- Openness  
  - overall: Mean 3.954 Std. Dev. 0.507 Min 1.575 Max 5.165 N =     535  
  - between: Mean 0.454 Std. Dev. 2.915 Min 5.059 n =      50  
  - within: Mean 0.261 Std. Dev. 1.505 Min 4.934 T-bar =    10.7

*Source: Annexes A–C, "_wp11149 - Annex A. Variables Definitions and Sources"*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2011/_wp11149.pdf_