## Macroeconomic Shocks and Conflict (WP/23/68)

## Source details

**Canonical URL:** [Macroeconomic Shocks and Conflict (WP/23/68)](https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023068-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2023/english/wpiea2023068-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2023/english/wpiea2023068-print-pdf.pdf.json)

---

### Executive summary and main quantitative findings
- Sample and identification:
  - Sample of 133 low- and middle-income countries over a 30-year period.
  - Uses the commodity Terms-of-Trade (ToT) index compiled by Gruss and Kebhaj (2019) as source of exogenous income variation.
  - Conflict data from the Uppsala Conflict Data Program Georeferenced Events Dataset (UCDP GED).
- Core empirical effects:
  - A negative commodity ToT shock leads to an increase in the number of conflict events and fatalities.
  - Quantitative estimates:
    - A negative shock equivalent to one percent of GDP leads to an increase of 0.05 conflict events per one million people.
    - The 0.05 events per one million people explains 1.6 percent of the average conflict incidences in the sample.
    - A negative ToT shock of one point of GDP is associated with an increase of 0.39 fatalities per one million people.
    - The 0.39 fatalities per one million people explains about 0.75 percent of the average fatality count in the sample.
  - Temporal pattern: effects peak within the first two years and decline thereafter, with persistence up to multiple years in vulnerable groups.
  - Spillovers: number of conflict events increases in neighboring countries in a statistically significant way within two to three years of the initial ToT shock.
- Heterogeneity:
  - Effects are roughly twice as large for Low-Income Countries (LICs) and Fragile and Conflict-affected States (FCS) relative to the sample average.
  - Larger impacts in countries with higher inequality and with limited fiscal capacity.
  - Among fiscal capacity measures, external debt is more relevant than fiscal balance and domestic debt for explaining vulnerability.
- Policy implication summary:
  - Evidence consistent with a fragility trap—macroeconomic shocks and conflict reinforcing each other.
  - Effective policies and well-tailored external financial support could help countries address the challenge.

### Data, measures, and transmission-channel framework
- Conflict measures and sample construction:
  - Dependent variable: number of conflict events per one million people (UCDP GED definition: organized violence incidents with at least one direct death).
  - Alternative dependent variable: number of fatalities per one million people.
  - Time coverage: 1989 to 2021; sample excludes high-income countries (World Bank GNI per capita classification above USD 12,535 in 2020).
  - Aggregation/cleaning: top one percentile of observations removed; conflict events normalized per one million people; missing values imputed with previous or subsequent year.
- Commodity ToT index construction and interpretation:
  - Commodity ToT index (Gruss and Kebhaj, 2019) built from 45 world commodity prices grouped into Energy, Metals, Food and beverages, Agricultural raw materials.
  - Weights: country-commodity price weighted by share in total net exports scaled to GDP using UN Comtrade trade data; three-year rolling averages; weights lagged to reflect international price changes.
  - One percentage point change in the index represents the equivalent percentage point change in GDP expected due to net commodity price changes.
  - Index normalized to 100 in 2012; between 10th and 90th percentiles the index fluctuates between a loss of about 10 percent of GDP and a gain of 8 percent of GDP.
- Conceptual transmission channels:
  1. Opportunity cost channel: negative export-price shocks lower household income and the opportunity cost of engaging in violence (proxies: Gini, unemployment, food insecurity).
  2. State capacity channel: lower national income reduces government revenue and capacity to prevent/mitigate conflict (proxies: fiscal deficits, external debt).
  3. Predation channel: higher commodity rents can increase conflict over resource capture (may offset other channels in some cases).
- Empirical identification:
  - Local projection method (Jorda, 2005) to estimate dynamic responses of conflict outcomes to ΔToT.
  - Estimation equation includes country fixed effects μ_{i,s}, time fixed effects λ_{t,s}, region-by-time fixed effects ν_{r,t,s}, lagged outcomes and lagged ToT terms.
  - Standard errors clustered by country; lag length chosen as two using Bayesian Information Criterion; unit root tests confirm stationarity.

### Descriptive statistics and geographic patterns
- Panel size and coverage:
  - Between 1989 and 2021, 4,412 observations for 133 countries.
- Key sample-level summary statistics (full sample unless noted):
  - Number of conflict events (per million population): Mean 3.19; Median 0.00; Standard Deviation 32.54; 10th percentile 0.00; 90th percentile 3.31.
  - Number of fatalities (per million population): Mean 51.74; Median 0.00; Standard Deviation 1392.17; 10th percentile 0.00; 90th percentile 32.82.
  - Commodity ToT index: Observations 4,072; Mean 100.23; Median 101.59; Standard Deviation 10.07; 10th percentile 90.65; 90th percentile 107.95.
  - GDP per capita (US$): Observations 4,196; Mean 2,710.29; Median 1,576.14; Standard Deviation 2,901.54; 10th percentile 331.82; 90th percentile 6,955.94.
  - Gini coefficient: Observations 4,026; Mean 41.39; Median 40.50; Standard Deviation 8.29; 10th percentile 31.70; 90th percentile 53.40.
  - Food insecurity (percent of population): Observations 2,574; Mean 37.82; Median 36.50; Standard Deviation 22.18; 10th percentile 11.20; 90th percentile 72.50.
  - External debt (percent of GDP): Observations 4,323; Mean 62.03; Median 42.56; Standard Deviation 77.63; 10th percentile -7.67; 90th percentile 119.70.
  - Fiscal balance (percent of GDP): Observations 4,521; Mean -2.77; Median -2.31; Standard Deviation 14.36; 10th percentile -37.59; 90th percentile 2.73.
- Geographic concentration (2021): conflict events mainly in Sub-Saharan Africa, the Middle East, and Central and Latin America; highest intensity in 2021: Afghanistan; countries with >50 conflict events per one million inhabitants in 2021 include Mexico, Syria, Yemen, Somalia, the Central African Republic, Cameroon, Mali.

### Main empirical results and heterogeneity
- Baseline effects:
  - A negative commodity ToT shock increases number of conflict events and fatalities, with effects peaking within the first two years and declining thereafter.
  - Estimates reiterated:
    - 1 percent of GDP negative ToT shock → +0.05 conflict events per one million people.
    - 1 percent of GDP negative ToT shock → +0.39 fatalities per one million people.
  - Interpretation: negative response implies dominance of opportunity-cost and state-capacity channels at the aggregate level.
- Conflict dyads (entry vs continuation):
  - Negative ToT shocks increase continuation of existing conflict dyads and prolong existing conflicts.
  - No significant effect on entry of new dyads—total increases driven by intensification and duration of ongoing conflicts.
- Income-level and fragility heterogeneity:
  - LICs and lower-MICs and PRGT-eligible countries: peak responses at 0.10 conflict events per one million people (almost double whole-sample effect); this 0.10 peak can explain 2.5 percent of average conflict incidents for these groups.
  - Responses in LICs/PRGT-eligible groups peak in the second year and persist for more than five years; higher-income groups show muted responses.
  - FCS (World Bank FCS list) display significantly larger responses; effect persists beyond 5 years.
- Transmission-channel evidence:
  - Opportunity-cost proxies: countries in the top 25th percentile of Gini, unemployment, or food insecurity experience larger responses (in the range 0.10 to 0.15 conflict events per one million people); low-inequality countries have estimates close to zero.
  - State-capacity proxies: countries with limited fiscal space—especially higher gross external debt—show substantially larger impacts.
  - Conclusion: evidence supports both opportunity-cost and state-capacity channels; predation channel may operate in specific resource-capture contexts but is de-emphasized by broad ToT index.
- Spillovers:
  - Distance-weighted index of other countries’ ToT with 1- to 2-year lags yields impulse responses significant about three years after initial shocks, indicating delayed spillovers to neighbors.
  - Mechanisms: cross-border expansion of violence, information effects, refugee flows (LICs and MICs host 74 percent of refugees), and negative economic spillovers via trade and regional linkages.

### Robustness and sensitivity analyses
- Alternative outcomes and IV:
  - Fatalities per one million people produce IRs similar to baseline but more persistent; a one-percentage point ToT change accounts for an additional 0.39 fatalities per one million people (presented elsewhere as 0.3 9 in source formatting), implying, with average population of 40 million, an increase of 15.6 fatalities in response to the initial shock and accounting for 0.75 percent of average sample fatalities.
  - Treating commodity ToT shocks as an instrument for per capita GDP growth yields results consistent with baseline.
- Sample-selection checks:
  - Excluding large commodity exporters (export share > 10 percent or > 20 percent for any commodity) leaves IRs mostly similar to baseline.
  - Excluding Middle East countries confirms baseline robustness.
- Controls and specification checks:
  - Baseline excludes macro/policy controls to allow transmission through these variables; including one-period lagged initial conditions (9 macro/policy, 1 social, 4 political instability, 1 polity score) yields impacts close to baseline with tighter confidence intervals.
  - Adding contemporaneous controls reduces initial-year IRs (90 percent interval near 0) but increases second-year IRs.
  - Wald test: adding macro, policy, and political variables does not improve overall explanatory power; forecasting performance with low-frequency data remains challenging.
- Interaction regressions:
  - Interaction terms with high Gini, high unemployment, high food insecurity, and high gross external debt are negative and significant in many specifications (significance levels reported as ***, **, * where indicated), consistent with amplified responses under high inequality and limited fiscal capacity.
  - External debt interactions often show the largest magnified coefficients.

### Policy implications and recommendations
- Strengthen fiscal capacity:
  - Adequate fiscal buffers and sustainable (external) debt levels are critical to mitigate ToT shock impacts and reduce likelihood of conflict escalation.
  - Fiscal buffers facilitate provision of social safety nets targeted to vulnerable households.
- Address inequality and inclusive growth:
  - Policies that increase incomes and reduce inequality enhance resilience to commodity shocks and lower the incentive to engage in violence via the opportunity-cost channel.
- External financial support and international engagement:
  - Well-tailored external financing and scaled-up humanitarian/development assistance can expand fiscal space, ease foreign-exchange constraints, and reduce conflict risks.
  - International institutions’ financing engagement can help reduce the likelihood of conflicts when countries face economic shocks.
- Monitor spillovers and regional dynamics:
  - Early monitoring of regional ToT shocks and refugee/information spillovers is important for preemptive policy responses and for supporting host countries bearing refugee burdens.

### Suggestions for future research
- Analyze different sources of macroeconomic shocks beyond the aggregate commodity ToT index; shock persistence heterogeneity may matter.
- Explore other forms of organized and unorganized violence (protests, riots) using granular datasets such as ACLED and the Conflict Barometer.
- Study effectiveness of policy and reform agendas in preventing conflicts and mitigating economic fallout; evaluate conditions and measures to overcome a "fragility trap."
- Investigate the role and effectiveness of external financial support to LICs and FCS in conflict prevention, containment, and durable post-conflict reconstruction.

*2023. Macroeconomic Shocks and Conflict. IMF Working Papers. WP/23/68.*

### 2023. Macroeconomic Shocks and Conflict. IMF Working Papers

### Macroeconomic Shocks and Conflict (WP/23/68)

### Executive Summary
- Sample and scope:
  - Sample of 133 low- and middle-income countries over a 30-year period.
  - Uses the commodity Terms-of-Trade (ToT) index compiled by Gruss and Kebhaj (2019) as source of exogenous income variation.
  - Conflict data from the Uppsala Conflict Data Program Georeferenced Events Dataset (UCDP GED).

- Main empirical findings:
  - A negative commodity ToT shock leads to an increase in the number of conflict events and fatalities.
  - Quantitative effects:
    - A negative shock equivalent to one percent of GDP leads to an increase of 0.05 conflict events per one million people.
    - The estimate of 0.05 events per one million people can explain 1.6 percent of the average conflict incidences in the sample.
    - A negative ToT shock of one point of GDP is associated with an increase of 0.39 fatalities per one million people.
    - The fatality estimate explains about 0.75 percent of the average fatality count in the sample.
  - Temporal pattern: the effect plays out over several years, with decreasing strength after the second year.
  - Heterogeneity:
    - Magnitude of effects is twice as large for Low-Income Countries (LICs) and Fragile and Conflict-affected States (FCS) compared with the sample average.
    - Larger impacts in countries with higher inequality and with limited fiscal capacity.
    - Among fiscal capacity measures, external debt is more relevant than fiscal balance and domestic debt for explaining vulnerability.

- Spillovers:
  - Number of conflict events increases in neighboring countries in a statistically significant way within two to three years of the initial ToT shock.

- Policy implication summary:
  - Results suggest the presence of a fragility trap—a vicious cycle of worsening economic conditions and deteriorating conflicts.
  - Effective policies and well-tailored external financial support could help countries address the challenge.

### Introduction and Motivation
- Context and motivation:
  - UCDP GED identifies conflict events as use of armed force by an organized actor against another organized actor, or against civilians that result in at least one direct death.
  - 17,000 conflicts occurred in 2021, claiming almost 120,000 deaths.
  - Recent global shocks with negative effects on output and aggregate income include the Covid-19 pandemic, severe climate events, and Russia’s invasion of Ukraine.
  - Ukraine war had strong impact on households’ purchasing power through energy, food, and fertilizer prices; LICs hit hardest due to large vulnerable populations and underdeveloped social safety nets.

- Research objective:
  - To analyze to what extent changes in a country’s commodity ToT can explain increases in the incidence and intensity of conflicts through effects on aggregate income.
  - To identify transmission channels from macroeconomic shocks to violence to inform policy design for preventing conflicts and achieving durable exits from fragility traps.

### Data, Transmission Channels, and Empirical Strategy
- Data sources and measures:
  - Commodity Terms-of-Trade (ToT) index: Gruss and Kebhaj (2019).
  - Conflict events and fatalities: UCDP GED (Sundberg and Melander, 2013).
  - Sample: 133 low- and middle-income countries, 30-year time span.
  - Alternative conflict measure used: number of fatalities to track intensity.

- Transmission channels (modeled via interactions with ToT shock):
  i. Opportunity cost channel:
     - Lower employment or lower pay could incentivize participation in violence.
     - Proxies: Gini coefficient (inequality), unemployment, share of population affected by food insecurity.
  ii. State capacity channel:
     - Lower national income reduces government revenue and capacity to prevent or mitigate conflict.
     - Proxies: fiscal deficits and external debt as shares of GDP.
  iii. Predation channel:
     - Higher commodity prices can increase conflict over distribution of rents, especially relevant for resource-rich sectors.

- Empirical method:
  - Local projection method (Jorda, 2005) to analyze dynamic effects of ToT changes on incidence and intensity of violent conflicts.
  - Interaction terms to explore transmission channels and heterogeneity (income groups, FCS status, inequality, fiscal capacity).
  - Robustness checks include alternative specifications and additional controls.

### Results
- Baseline specification:
  - Negative commodity ToT shock increases number of conflict events and fatalities, with effects peaking within the first two years and declining thereafter.
  - Exact quantitative results reiterated:
    - 1 percent of GDP negative ToT shock → +0.05 conflict events per one million people.
    - 1 percent of GDP negative ToT shock → +0.39 fatalities per one million people.

- Heterogeneity and channels:
  - LICs and FCS: effects twice as large relative to sample average.
  - Inequality: higher inequality amplifies the conflict response (supports opportunity cost channel).
  - Fiscal capacity: limited fiscal space amplifies the conflict response (supports state capacity channel); external debt is the most relevant fiscal indicator in this context.
  - Predation channel discussed as potential countervailing mechanism in some contexts (e.g., capital-intensive commodity revenue capture).

- Evolution of conflict dyads and entry/exit:
  - Responses of continuation and entry of conflict dyads documented, indicating persistence dynamics in conflicts following shocks.

- Spillover analysis:
  - Significant spillover effects: neighboring countries experience increases in conflict events within two to three years of an initial ToT shock.

### Robustness Checks
- Robustness exercises:
  - Results are robust to a number of plausible variations in model specification.
  - Alternative specifications and additional controls were tested (figures and tables cited show sensitivity analyses).

- Additional findings in robustness:
  - The findings on heterogeneity (LICs, FCS, inequality, fiscal constraints) remain supported across robustness checks.
  - Spillover patterns persist under alternative model specifications.

### Concluding Remarks
- Summary of implications:
  - The empirical evidence supports a causal link from adverse commodity ToT shocks to increases in conflict incidence and intensity.
  - Amplified effects in poorer, more unequal, and fiscally constrained countries point to policy-relevant vulnerability factors.
  - The combination of these dynamics suggests the risk of a fragility trap, with macroeconomic shocks and conflict reinforcing each other.

- Policy takeaways:
  - Strengthening fiscal capacity and social safety nets could mitigate the impact of commodity shocks on conflict.
  - Well-tailored external financial support and policies that address inequality can help reduce vulnerability to conflict escalation following macroeconomic shocks.
  - Monitoring spillovers and regional dynamics is important for preemptive policy responses.

*2023. Macroeconomic Shocks and Conflict. IMF Working Papers. WP/23/68.*

### 3.1   Data

### 3.1 Data

### Conflict data: scope, definitions, and sample construction
- Dependent variable: intensity of conflict measured as the number of conflict events per one million people.
- Data source: Uppsala Conflict Data Program Georeferenced Events Dataset (UCDP GED) (Sundberg and Melander, 2013).
- Definition of conflict events: organized violence incidents involving use of armed force by an organized actor against another organized actor, or against civilians, resulting in at least one direct death.
- Types of conflict events included:
  - state-based conflicts (incidents between two parties where at least one is the government of a state);
  - non-state-based conflicts (between two parties neither of which is the government of a state);
  - one-sided conflicts (armed force used by the government of a state or by a formally organized group against civilians).
- Excluded: non-organized or non-violent incidents such as protests, riots, and violence by civilians.
- UCDP GED coverage: events for all dyads and actors that have surpassed the 25 deaths threshold at least in one calendar year; information based on global news reports via the Dow Jones Factiva aggregator.
- Time coverage: 1989 to 2021.
- Sample: 133 countries, excluding high-income countries according to the World Bank’s GNI per capita classification (above USD 12,535 in 2020). (Full country list in Annex I of source.)
- Aggregation and cleaning:
  - Number of conflict events aggregated by country and year.
  - Top one percentile of observations of each variable removed as outliers.
  - Conflict events normalized per one million people to account for country size.
  - Missing values in other variables imputed using the previous or subsequent year’s observations to maintain sample size.
- Alternative dependent variable: number of fatalities per one million people; analysis obtains similar results.
- Additional analyses: entry and continuation of conflict dyads investigated to understand stages of conflict formation (details in Section IV).

### Descriptive snapshot and geographic patterns (2021)
- Geographic concentration (2021): conflict events mostly concentrated in Sub-Saharan Africa, the Middle East and Central and Latin America.
- Highest intensity in 2021: Afghanistan.
- Countries recording more than 50 conflict events per one million inhabitants in 2021: Mexico, Syria, Yemen, Somalia, the Central African Republic, Cameroon, Mali.

### Commodity terms-of-trade (ToT) index: identification strategy and construction
- Identification challenge: control for endogeneity because macroeconomic conditions are affected by violence.
- Exogenous variation used: country-specific commodity ToT index constructed by Gruss and Kebhaj (2019) and maintained by the IMF Research Department.
- Coverage: 182 economies in the commodity ToT dataset.
- Purpose: provides estimates for gains and losses in income associated with changes in international commodity prices; a one percentage point change in the index represents the equivalent percentage point change in GDP expected for each country due to net commodity price changes.
- Commodity price inputs: 45 world commodity prices from the IMF Primary Commodity Price database, classified into four categories:
  1. Energy: coal, crude oil, natural gas.
  2. Metals: aluminum, copper, gold, iron ore, lead, nickel, tin, uranium, zinc.
  3. Food and beverages: bananas, barley, beef, cocoa, coffee, corn, fish, fish meal, groundnuts, lamb, olive oil, oranges, palm oil, poultry, rapeseed oil, rice, shrimp, soybean meal, soybean oil, soybeans, sugar, sunflower seed oil, swine meat, tea, wheat.
  4. Agricultural raw materials: cotton, hard logs, hard sawnwood, hides, natural rubber, soft logs, soft sawnwood, wool.
- Weighting scheme:
  - Each commodity price for every country is weighted by its share in total net exports scaled to GDP to proxy impact on aggregate income.
  - Country-commodity level trade data sourced from the United Nations Comtrade Database.
  - Shares built with three-year rolling averages for trade values; weights are lagged so index reflects international price changes rather than endogenous trade-volume responses.
- Distinction: commodity ToT index differs from a standard ToT measure by accounting for the share of each export and import good in a country’s economic structure; Gruss and Kebhaj (2019) find only the commodity ToT index influences macroeconomic aggregates statistically significantly.
- Cross-country variation example (2021): within Middle East and North Africa, Qatar, Kuwait and Libya experienced positive ToT changes, while Jordan and Djibouti suffered significant losses; changes vary even across oil exporters due to different net export shares relative to GDP and impacts of other commodities.
- Figure 4 in source: commodity ToT index year-on-year change (percent) for sample countries with highest gains and losses in 2021.

### Descriptive statistics (sample-level)
- Panel observations and period:
  - Between 1989 and 2021, there are 4,412 observations for 133 countries.
- Key sample averages (full sample unless noted):
  - Number of conflict events (per million population): Mean 3.19; Median 0.00; Standard Deviation 32.54; 10th percentile 0.00; 90th percentile 3.31.
  - Number of fatalities (per million population): Mean 51.74; Median 0.00; Standard Deviation 1392.17; 10th percentile 0.00; 90th percentile 32.82.
  - Number of new conflict dyads: Observations 4,765; Mean 0.90; Median 0.00; Standard Deviation 2.22; 10th percentile 0.00; 90th percentile 3.00.
  - Number of continuing conflict dyads: Observations 3,232; Mean 1.08; Median 0.00; Standard Deviation 2.45; 10th percentile 0.00; 90th percentile 3.00.
  - Commodity ToT index: Observations 4,072; Mean 100.23; Median 101.59; Standard Deviation 10.07; 10th percentile 90.65; 90th percentile 107.95.
- Income-group differences:
  - Low-Income and Lower Middle-Income Countries (GDP per capita below US$4,045 in 2020):
    - Observations for conflict events 2,696; Number of conflict events (per million population): Mean 3.98; Median 0.00; Standard Deviation 34.78; 10th percentile 0.00; 90th percentile 4.95.
    - Number of fatalities (per million population): Mean 75.06; Median 0.00; Standard Deviation 1774.51; 10th percentile 0.00; 90th percentile 56.70.
    - Number of new conflict dyads: Observations 3,049; Mean 1.17; Median 0.00; Standard Deviation 2.58; 10th percentile 0.00; 90th percentile 4.00.
    - Number of continuing conflict dyads: Observations 2,336; Mean 1.21; Median 0.00; Standard Deviation 2.69; 10th percentile 0.00; 90th percentile 3.00.
    - Commodity ToT index: Observations 2,585; Mean 100.33; Median 101.66; Standard Deviation 9.27; 10th percentile 91.55; 90th percentile 107.71.
  - Upper Middle-Income Countries:
    - Observations for conflict events 1,749; Number of conflict events (per million population): Mean 1.95; Median 0.00; Standard Deviation 28.38; 10th percentile 0.00; 90th percentile 1.67.
    - Number of fatalities (per million population): Mean 14.83; Median 0.00; Standard Deviation 183.87; 10th percentile 0.00; 90th percentile 4.60.
    - Number of new conflict dyads: Observations 2,092; Mean 0.36; Median 0.00; Standard Deviation 1.14; 10th percentile 0.00; 90th percentile 1.00.
    - Number of continuing conflict dyads: Observations 992; Mean 0.70; Median 0.00; Standard Deviation 1.56; 10th percentile 0.00; 90th percentile 2.00.
    - Commodity ToT index: Observations 1,611; Mean 99.87; Median 101.28; Standard Deviation 11.33; 10th percentile 88.04; 90th percentile 108.42.
- Other economic variables (sample means and dispersion):
  - GDP per capita (US$): Observations 4,196; Mean 2,710.29; Median 1,576.14; Standard Deviation 2,901.54; 10th percentile 331.82; 90th percentile 6,955.94.
  - Gini coefficient: Observations 4,026; Mean 41.39; Median 40.50; Standard Deviation 8.29; 10th percentile 31.70; 90th percentile 53.40.
  - Unemployment rate (percent): Observations 4,290; Mean 8.75; Median 6.90; Standard Deviation 7.13; 10th percentile 1.80; 90th percentile 19.03.
  - Food insecurity (percent of population): Observations 2,574; Mean 37.82; Median 36.50; Standard Deviation 22.18; 10th percentile 11.20; 90th percentile 72.50.
  - External debt (percent of GDP): Observations 4,323; Mean 62.03; Median 42.56; Standard Deviation 77.63; 10th percentile -7.67; 90th percentile 119.70.
  - Fiscal balance (percent of GDP): Observations 4,521; Mean -2.77; Median -2.31; Standard Deviation 14.36; 10th percentile -37.59; 90th percentile 2.73.
- Index normalization and percentile range:
  - Commodity ToT index normalized to 100 in 2012.
  - Between the tenth and ninetieth percentiles the index fluctuates between a loss of about 10 percent of GDP and a gain of 8 percent of GDP.
- Note on high dispersion: standard deviations of conflict variables are high reflecting a few extremely severe conflict episodes; top 1% excluded in regression analysis.

### Pairwise correlations and relationships
- Summary correlations (Table 2 highlights):
  - Different conflict measures positively correlated with each other.
  - Commodity ToT index negatively correlated with conflict variables (suggesting deterioration associated with higher conflict intensity).
  - Conflict variables negatively correlated with GDP per capita.
  - Food insecurity and external debt mostly positively correlated with measures of conflict.
  - Gini coefficient and unemployment rate are negatively correlated with all measures of conflict in the pairwise table.
  - Many indicators are correlated with GDP per capita, motivating rigorous econometric identification.
- Commodity price index correlations (Table 3):
  - Correlations among sub-category indexes (Food and Beverages, Raw Agricultural Materials, Base Metals, Energy, Crude Oil) are moderate to high (e.g., Energy and Crude Oil index correlation 0.971).
  - Correlations between the ToT index and sub-category indexes are quite low (ToT index correlations: Food and Beverages 0.03; Raw Agricultural Materials 0.0697; Base Metals 0.0777; Energy 0.0404; Crude Oil 0.0412).
  - ToT index is highly correlated with Commodity Export price index (0.8074) and negatively correlated with Commodity Import price index (-0.28491).
  - Note: table shows correlation of year-on-year changes of indexes.

### Transmission channels (summary of conceptual framework)
- Three channels by which income shocks from commodity ToT changes may affect incidence and intensity of conflicts:
  1. Opportunity costs:
     - Negative commodity export price shock lowers household income, reducing opportunity cost of engaging in conflict.
     - Example: decline in coffee and cocoa prices in Côte d’Ivoire in the 1980s and 1990s contributed to competition over land and rising tensions.
     - Commodity import price increases also reduce real household income (e.g., food price increases in 2007–2008 linked to food riots).
  2. State capacity:
     - Negative export price shocks lower government revenue; import price increases raise fiscal costs of subsidies and support measures, weakening state capacity to mitigate shocks.
     - Example: Nigeria’s oil revenue fall in 2014–2015 constrained government resources to address Boko Haram expansion.
  3. Predation:
     - Negative ToT shock from lower export prices lowers gains from rent capture, thereby potentially decreasing incentives for conflict through predation (often discussed for oil and other natural resource sectors).
- Net effect: overall impact of commodity price changes on conflicts is ambiguous ex ante and requires careful regression analysis.

### Empirical strategy (estimation framework)
- Technique: local projections (Jorda, 2005) used to estimate impact of a change in the commodity ToT index on change in outcome variables (baseline: number of conflict events per one million people).
- Local projection setup:
  - Impulse response function defined as difference in expected future outcomes conditional on shock vs. no shock.
  - Estimation equation (as in source):
    - y_{i,t+s} - y_{i,t-1} = β_s ΔToT_{i,t} + Σ_{l=1}^L α_{l,s} Δy_{i,t-l} + Σ_{k=1}^K δ_{k,s} ΔToT_{i,t-k} + μ_{i,s} + λ_{t,s} + ν_{r,t,s} + ε_{i,t,s}, for s = 0,1,2,...
  - Where ΔToT_{i,t} is percentage change in commodity ToT index for country i at time t.
  - Fixed effects:
    - μ_{i,s}: country fixed effects (time-invariant country characteristics).
    - λ_{t,s}: time fixed effects (global common components).
    - ν_{r,t,s}: region-by-time fixed effects (time-varying regional shocks).
  - Standard errors clustered by country.
  - Lag structure: include lagged outcome and explanatory variables to address serial correlation; choose lag length of two using the Bayesian Information Criterion in the Levin-Lin-Chu panel unit root test.
  - Unit root test: confirms explanatory and outcome variables are stationary.
- Treatment of ΔToT as exogenous shock in baseline; robustness tests of this assumption presented in Section V of source.
- Instrumental approach: commodity ToT index is used as explanatory variable in baseline (following convention); also tested as an instrument yielding similar results (additional results presented in source).

*Source: wpiea2023068-print-pdf — Section 3.1 Data.*

### Section V.

### Section V.

### IV. Results

- 4.1 Baseline Specification
  - A deterioration of the commodity ToT increases the frequency of conflicts.
  - A negative ToT shock equivalent to one percent of GDP increases the likelihood of experiencing a conflict by 0.05 events per one million people in the year following the shock.
  - The estimate of 0.05 events per one million people is equivalent to or explains 1.6 percent of the average country’s conflict events per year in the sample.
  - Conflicts remain more likely for 2-3 years after a shock before the impact gradually decays.
  - The sign of estimated coefficients is flipped and multiplied by 100 in presented figures; shaded areas indicate the 68 and 90 percentiles of the responses.
  - Interpretation: The negative response implies dominance of the opportunity-cost and state-capacity channels at the aggregate level; use of a broad-based commodity ToT index de-emphasizes idiosyncratic sectoral predation effects.

- 4.2 Evolution of Conflict dyads
  - Conflict dyads decomposed into “entry” (active this year but not previous year) and “continuation” (active this year and previous year).
  - Negative commodity ToT shocks:
    - Increase the number of continued conflict dyads and prolong existing conflicts (Panel (a) results).
    - Do not significantly trigger entry of new dyads (Panel (b) results).
  - Implication: Increase in total number of conflict events is attributable to intensification by existing dyads; economic factors are important for duration/intensity rather than initial onset.

- 4.3 Countries at Risks
  - Income level heterogeneity (interaction regression as in equation (3)):
    - Sample split: LICs and lower MICs (GNI per capita below US$4,045 in 2020) versus upper MICs; complementary classification: PRGT eligibility.
    - Lower-income groups include 81 countries (World Bank) and 69 PRGT-eligible countries.
  - Figure 8 findings:
    - Responses in LICs/lower-MICs and PRGT-eligible countries peak at 0.10 conflict events per one million people — almost double the whole-sample effect.
    - The 0.10 peak can explain 2.5 percent of the average conflict incidents in these country groups.
    - Impact is highest in the second year after the initial shock and is persistent for more than five years; first-year response close to that of higher-income groups.
    - Higher-income groups show muted responses over the estimation period.
  - FCS (Fragile and Conflict-affected Situations):
    - Using World Bank FCS list (37 economies in 2023; most are LICs), applying 2006 classification for earlier years.
    - FCS display a significantly larger number of conflict events in response to a ToT shock; initial impact close to LICs and effect persists beyond 5 years.

- 4.4 Tracking the Transmission Channels
  - Aim: Disentangle opportunity-cost and state-capacity channels using proxy country characteristics in the interaction specification.
  - Inequality measures (World Bank): (i) Gini coefficient, (ii) unemployment rate, (iii) share of population exposed to food insecurity.
    - Sample split at top 25th percentile for each measure; variables lagged by one period from the shock.
    - Countries in the highest quantile for each inequality measure have considerably larger responses: increases in the likelihood of conflicts in the range of 0.10 to 0.15 conflict events per one million people.
    - Estimates for low-inequality countries are close to zero.
    - Interpretation: Supports the opportunity-cost hypothesis; consistent with economics of crime (Becker, 1968) logic.
  - Fiscal capacity measures: gross external debt and fiscal balance (percent of GDP) from World Bank and July 2022 WEO vintage.
    - Split at third quartile cutoff (top 25th percentile) to identify limited fiscal space.
    - Countries with relatively low fiscal space show larger estimated impacts; coefficient substantially larger for countries with higher external debt (though standard error is somewhat large).
    - Interpretation: Limited fiscal capacity constrains government response to negative ToT shocks, supporting the state-capacity channel.
  - Overall: Evidence for presence of both opportunity-cost and state-capacity channels; proxies may also capture deep-rooted structural issues (e.g., past conflict intensity affecting fiscal capacity).

- 4.5 Testing for Spillovers Across Countries
  - Constructed distance-weighted index of other countries’ commodity ToT using distances from United States International Trade Commission and adding 1- to 2-year lagged terms (m = 1, 2) to baseline regression.
  - Results (Figure 11): Impulse responses become significant three years after the initial ToT shocks, indicating spillovers occur with a time lag.
  - Possible spillover mechanisms discussed:
    - Direct expansion of violent activities across borders.
    - Information effects influencing beliefs and triggering conflict elsewhere.
    - Refugee flows weighing on host-country budgets and weakening fiscal capacity (LICs and MICs host 74 percent of refugees).
    - Negative economic spillovers via trade and regional economic linkages.
  - Implication: Economic shocks can have conflict impacts beyond initial countries hit; state-capacity support and scaled-up humanitarian/development assistance are critical.

### V. Robustness Check

- 5.1 Alternative Specifications
  - Alternative outcome: fatalities per one million people (UCDP GED).
    - IRs similar to baseline but more persistent; impact remains significant for three years after the shock.
    - Estimated effect: a one-percentage point change in the commodity ToT accounts for an additional 0.3 9 fatalities per one million people.
    - With average population of 40 million among sample countries, the estimate implies an increase of 15.6 fatalities as a response to the initial shock.
    - This can account for 0.75 percent of the average fatalities for the whole sample.
  - Commodity ToT shocks as an instrumental variable (IV) for per capita GDP growth:
    - IV result consistent with baseline.
    - Baseline uses commodity ToT shocks directly to capture channels beyond aggregate income (e.g., resource competition) but results imply main channel is through aggregate income.
  - Sample selection robustness:
    - Excluding large commodity exporters: computed export share for each commodity-country pair using UN Comtrade 2018; excluded countries with export share > 10 percent for any commodity — leading to exclusion of 36 countries (listed in Annex I).
      - IRs remain mostly similar to baseline.
    - Robustness also holds when using 20 percent export share cutoff (excludes 24 countries).
    - Excluding Middle East countries also confirms robustness of baseline result.

*Source: wpiea2023068-print-pdf - Section V.*

### 5.2   Additional Controls

### 5.2   Additional Controls

### Sensitivity to control variables
- Baseline specification excludes macroeconomic or policy controls to allow transmissions of an initial shock through these variables; specification relies on the premise that the commodity ToT shocks are not affected by these variables.
- Panel (a) of Figure 13: include one-period lagged initial conditions (to avoid endogeneity):
  - 9 macroeconomic and policy variables: per capita GDP growth, unemployment rate, CPI inflation, current account balance, fiscal balance, gross external debt, foreign reserve, USD-local currency exchange rate, and broad money.
  - 1 social indicator: Gini coefficient.
  - 4 indicators of political instability: the number of major cabinet changes, the number of successful, attempted, and planned coups.
  - 1 indicator of political regime: polity score.
- Result: estimated impact of the commodity ToT shock is close to the baseline estimation, confirming exogeneity of commodity ToT shocks with respect to initial conditions.
- Tighter confidence intervals in Panel (a) suggest efficiency gains from removing variation of control variables from error terms.
- Panel (b) of Figure 13: add contemporaneous terms of the same control variables (identifying the coefficient of commodity ToT as marginal impact keeping these controls constant in the initial period).
  - Result: impulse responses (IRs) become smaller in the initial year with the 90 percent interval almost covering 0, whereas they increase in the second year.
- Overall: controlling various variables only modestly affects baseline results.

### Interaction with income groups and presence of conflict (Table 4 results)
- Purpose: address concern that inequality and fiscal capacity may reflect deeper structural factors by adding interaction terms with income groups or a dummy for presence of conflict events.
- Key messages from Table 4:
  - Interaction terms with measures of inequality and fiscal capacity remain significant in most specifications, confirming the opportunity-cost and fiscal capacity channels.
  - External debt (gross external debt) appears highly relevant for representing fiscal space in many LICs and MICs.
  - Interaction terms with income group and conflict events are significant in most specifications, indicating mean income level and inequality affect the impact of income changes on conflict.

- Panel (a) Income level — Dependent variable: Number of conflict events per capita
  - Δlog(Commodity ToT):
    - Column (1): 2.382 (1.835)
    - Column (2): 2.163 (3. 562)
    - Column (3): -3.037 (2.097)
    - Column (4): 1.415 (1.849)
    - Column (6): 1.816 (2.693)
  - Δlog(Commodity ToT)*1(LICs and lower-MICs):
    - Column (1): -7.610** (2.990)
    - Column (2): -5.076 (3.782)
    - Column (3): 0.590 (2.874)
    - Column (4): -4.973* (2.672)
    - Column (6): -8.171*** (3.070)
  - Δlog(Commodity ToT)*1(High Gini Coef):
    - Column (1): -8.983*** (2.677)
  - Δlog(Commodity ToT)*1(High unemployment rate):
    - Column (2): -12.478*** (4.583)
  - Δlog(Commodity ToT)*1(High food Insecurity):
    - Column (3): -11.981*** (3.406)
  - Δlog(Commodity ToT)*1(High gross external debt):
    - Column (4): -19.498*** (5.793)
  - Δlog(Commodity ToT)*1(Low Fiscal Balance):
    - Column (6): -10.462 (7.847)
  - Observations: 3,263; 3,563; 2,107; 3,361; 3,420 (respectively for columns reported).
  - R-squared: 0.210; 0.172; 0.270; 0.227; 0.222 (respectively).
  - Year FE: Yes; Country FE: Yes; Region-by-year FE: Yes; SE clustering: Country.
  - Significance notation: *** p<0.01, ** p<0.05, * p<0.1.

- Panel (b) Presence of conflict — Dependent variable: Number of conflict events per capita
  - Δlog(Commodity ToT):
    - Column (1): 1.342 (1.343)
    - Column (2): 3.642 (2.221)
    - Column (3): 3.045 (1.886)
    - Column (4): 3.852* (2.089)
    - Column (6): 0.935 (1.702)
  - Δlog(Commodity ToT)*1(Number of conflict event>0):
    - Column (1): -5.226 (3.855)
    - Column (2): -8.032* (4.138)
    - Column (3): -10.117*** (3.226)
    - Column (4): -10.593*** (3.054)
    - Column (6): -7.317** (3.296)
  - Δlog(Commodity ToT)*1(High Gini Coef):
    - Column (1): -10.637*** (2.612)
  - Δlog(Commodity ToT)*1(High unemployment rate):
    - Column (2): -12.207*** (4.409)
  - Δlog(Commodity ToT)*1(High food Insecurity):
    - Column (3): -9.714*** (3.602)
  - Δlog(Commodity ToT)*1(High gross external debt):
    - Column (4): -18.343*** (5.610)
  - Δlog(Commodity ToT)*1(Low Fiscal Balance):
    - Column (6): -9.376 (7.336)
  - Observations: 3,288; 3,589; 2,107; 3,386; 3,445 (respectively).
  - R-squared: 0.207; 0.172; 0.271; 0.227; 0.220 (respectively).
  - Year FE: Yes; Country FE: Yes; Region-by-year FE: Yes; SE clustering: Country.
  - Notes: The dependent variable is one period ahead other than in column (2) where the contemporaneous variable is used. One- and two-period lagged dependent and independent variables are included. Timing chosen as the period when the impact is peaked.

### Model specification and forecasting performance
- A Wald test indicates inclusion of the macro, policy, and political variables shown in Figure 13 does not improve overall explanatory power of regressions.
- Implication: stylized statistical models based on conventional macroeconomic variables face challenges in predicting conflict incidence and intensity, consistent with previous studies reporting relatively low forecasting performance with low frequency data.

### Concluding implications and policy recommendations
- Main empirical conclusions:
  - A negative commodity ToT shock significantly increases intensity of conflicts (measured by number of conflict events per one million population).
  - Impact tends to be larger and more persistent for Low-Income Countries (LICs) and Fragile and Conflict-affected States (FCS).
  - Impact is positively correlated with level of inequality and a country’s external debt burden.
  - Second-round effects: ToT shocks generate spillovers to neighboring countries’ security through conflict spillovers.
  - Heterogeneous consequences: impact depends on underlying macroeconomic, institutional, and geographical conditions.
- Policy implications explicitly stated:
  - Inclusive growth that increases incomes and reduces inequality should enhance resilience to shocks and help prevent violence.
  - Adequate fiscal buffers in the form of sustainable (external) debt levels are critical to mitigate ToT shock impacts and the likelihood of conflicts.
  - Fiscal buffers can facilitate provision of adequate social safety nets, ideally targeted to vulnerable households.
  - Financing engagement by international institutions can help countries reduce likelihood of conflicts when faced with economic shocks by increasing scope for budget spending and easing foreign exchange constraints.

### Suggestions for future research (as listed)
- Analyze different sources of macroeconomic shocks beyond the overall commodity ToT index; differences in shock persistence may have different implications.
- Explore different types of organized violence and other forms of social unrest (protests, riots, non-organized violence); use granular data sources such as ACLED and the Conflict Barometer.
- Study effectiveness of policy and reform agendas in preventing conflicts and/or mitigating their economic fallout; open questions remain on measures policymakers can take to build resilience.
- Investigate the role of external financial support to LICs and FCS in conflict prevention, containment, and durable post-conflict reconstruction.
- Examine conditions conducive to overcoming a "fragility trap" — the nexus between weak economic performance and conflicts — including effective policy choices and adequate external financing support.

*Source: wpiea2023068-print-pdf — Section 5.2 Additional Controls.*

### References

### References

### Themes in the referenced literature
- Poverty, political economy, and terrorism:  
  - Abadie, A. (2006). Poverty, Political Freedom, and the Roots of Terrorism. American Economic Review, 96(2):50–56.  
  - Krueger, A.B. and J. Malečková. (2003). Education, Poverty and Terrorism: Is There a Causal Connection? Journal of Economic Perspectives, 17 (4):119–144.
- Natural resources, commodity prices, and conflict:  
  - Auty, R. M. 1993. Sustaining Development in the Mineral Economies: The Resource Curse Thesis. London: Routledge.  
  - Berman, N., M. Couttenier, D. Rohner, and M. Thoenig. (2017). This Mine Is Mine! How Minerals Fuel Conflicts in Africa. American Economic Review, 107 (6): 1564–1610.  
  - Brückner, M. and A. Ciccone. (2010). International Commodity Prices, Growth and the Outbreak of Civil War in Sub‐Saharan Africa. The Economic Journal, 120(544):519–534.  
  - Dube, O. and J.F. Vargas (2013). Commodity Price Shocks and Civil Conflict: Evidence from Colombia. The Review of Economic Studies, 80(4 (285)):1384–1421.  
  - Bazzi, S. and C. Blattman. (2014). Economic Shocks and Conflict: Evidence from Commodity Prices. American Economic Journal: Macroeconomics, 6(4): 1 –38.  
  - Cotet, A. M. and K. K. Tsui. (2013). Oil and conflict: What Does the Cross-Country Evidence Really Show? American economic journal: Macroeconomics, 5(1):49–80.
- Climate, weather, and conflict/economic production:  
  - Burke, M., S.M. Hsiang, and E. Miguel. (2015). Global Non-Linear Effect of Temperature on Economic Production. Nature, 527:235–239.  
  - Hsiang, S.M., Burke, M., and Miguel, E. (2013). Quantifying the Influence of Climate on Human Conflict. Science, 341, 1235367  
  - Diallo, Y. and R. Tapsoba. (2022). Climate Shocks and Domestic Conflicts in Africa. IMF Working Paper WP/22/250, International Monetary Fund.  
  - Harari, M. and E. La Ferrara. (2018). Conflict, Climate, and Cells: A Disaggregated Analysis. The Review of Economics and Statistics,100 (4): 594–608.  
  - Sarsons, H. (2015). Rainfall and Conflict: A Cautionary Tale. Journal of Development Economics, 115(July):62–72.
- Geography, spatial dependence, and diffusion of conflict:  
  - Anselin L., and J. O’Loughlin. (1992). Geography of International Conflict and Cooperation: Spatial Dependence and Regional Context in Africa. In New Politics. Routledge.  
  - Buhaug, H. and S. Gates. (2002). The Geography of Civil War. Journal of Peace Research, 39(4):417–433.  
  - Buhaug, H. and K. S. Gleditsch. (2008). Contagion or Confusion? Why Conflicts Cluster in Space. International Studies Quarterly, 52(2): 215–233  
  - Salehyan, I. and K. Gleditsch. (2006). Refugees and the Spread of Civil War. International Organization, 60(2):335–366.  
  - Sundberg, R. and E. Melander. (2013). Introducing the UCDP Georeferenced Event Dataset. Journal of Peace Research, 50(4), 523–532.
- Civil war, insurgency theory, and state capacity:  
  - Besley, T.J. and T. Persson. (2010). State Capacity, Conflict, and Development. Econometrica, 78(1):1–34.  
  - Blattman, C. and E. Miguel. (2010). Civil War. Journal of Economic Literature, 48(1):3–57.  
  - Collier, P. and A. Hoeffler. (2004). Greed and Grievance in Civil War. Oxford Economic Papers, 56(4):563–595.  
  - Fearon, J. D., and D. D. Laitin (2003). Ethnicity, Insurgency, and Civil War. The American Political Science Review, 97(1):75–90.  
  - Elbadawi, I. and N. Sambanis. (2002). How Much War Will We See? Explaining the Prevalence of Civil War. The Journal of Conflict Resolution, 46(3):307–334.
- Economic shocks, growth collapses, and macroeconomic impacts of conflict:  
  - Cerra, V. and S. C. Saxena. (2008). Growth Dynamics: The Myth of Economic Recovery. American Economic Review, 98(1):439–457.  
  - Rodrik, D. (1999). Where Did All the Growth Go? External Shocks, Social Conflict, and Growth Collapses. Journal of Economic Growth, 4(4):385–412.  
  - Novta, N. and E. Pugacheva. (2021). The Macroeconomic Costs of Conflict. Journal of Macroeconomics, 68, 103286.  
  - Hadzi-Vaskov, M., S. Pienknagura, and L.A. Ricci. (2021). The Macroeconomic Impact of Social Unrest. IMF Working Paper WP/21/135, International Monetary Fund.  
  - Rother, B., G. Pierre, D. Lombardo, R. Herrala, P. Toffano, E. Roos, G. Auclair, K. Manasseh, V. Gaspar, M. Obstfeld, et al. (2016). The Economic Impact of Conflicts and the Refugee Crisis in the Middle East and North Africa. IMF Staff Discussion Note, SDN/16/08.
- Prediction, early warning, and modeling of conflict:  
  - Cederman, L. and N. B. Weidmann. (2017). Predicting Armed Conflict: Time to Adjust Our Expectations? Science, 355, 474–476.  
  - Kleinberg, J., J. Ludwig, S. Mullainathan, and Z. Obermeyer. (2015). Prediction Policy Problems. American Economic Review: Papers and Proceedings, 105(5):491–495.  
  - Mueller, H., C. Rauh, and A. Ruggieri. (2022). Dynamic Early Warning and Action Model. BSE Working Paper 1355.  
  - Jorda, O. (2005). Estimation and Inference of Impulse Responses by Local Projections. American Economic Review, 95(1):161–182.  
  - Olea, M., J. Luis, and M. Plagborg-Møller. (2021). Local Projection Inference is Simpler and More Robust Than You Think. Econometrica, 89(4):1789–1823.
- Food prices, food security, and social unrest:  
  - Bellemare, M. F. (2015). Rising Food Prices, Food Price Volatility, and Social Unrest. American Journal of Agricultural Economics, 97(1):1–21.  
  - Fjelde, H. (2015). Farming or Fighting? Agricultural Price Shocks and Civil War in Africa. World Development, 67:525–534.  
  - Schneider, M. (2008). “We Are Hungry!” – A Summary Report of Food Riots, Government Responses, and States of Democracy in 2008. Working Paper, Department of Development Sociology, Cornell University.  
  - van Weezel, S. (2016). Food Imports, International Prices, and Violence in Africa. Oxford Economic Papers, 68(3):758–781.  
  - Rother, B., S. Sosa, L. Kohler, G. Pierre, and others. (2022). Tackling the Global Food Crisis: Impact, Policy Response, and the Role of the IMF? IMF Note 2022/004, International Monetary Fund, Washington, DC.
- Media, information diffusion, and mobilization:  
  - Khondker, H. H. (2011). Role of the New Media in the Arab Spring. Globalizations, 8(5):675–679.  
  - Salvatore, A. (2011). Before (and After) the “Arab Spring”: From Connectedness to Mobilization in the Public Sphere. Oriente Moderno, 91(1): 5–12.  
  - Zhang, B. (2022). Newspaper, Post Office, and Protest: How Do Political Information Diffusion and Social Interactions Affect Collective Action in Late Imperial China? Mimeo.

### Methodological and data resources cited
- Datasets and georeferenced event data:  
  - Sundberg, R. and E. Melander. (2013). Introducing the UCDP Georeferenced Event Dataset. Journal of Peace Research, 50(4), 523–532.  
- IMF working papers, notes, and departmental papers referenced by number:  
  - Akanbi, O., N. Gueorguiev, J. Honda, P. Mehta, K. Moriyama, K. Primus, and M. Sy. (2021). Avoid a Fall or Fly Again: Turning Points of State Fragility. IMF Working Paper WP/21/133, International Monetary Fund.  
  - Diallo, Y. and R. Tapsoba. (2022). Climate Shocks and Domestic Conflicts in Africa. IMF Working Paper WP/22/250, International Monetary Fund.  
  - Gruss, B. and S. Kebhaj. (2019). Commodity terms of trade: A new database. IMF Working Paper WP/19/21, International Monetary Fund.  
  - Hadzi-Vaskov, M., S. Pienknagura, and L.A. Ricci. (2021). The Macroeconomic Impact of Social Unrest. IMF Working Paper WP/21/135, International Monetary Fund.  
  - International Monetary Fund. (2022b). Regional Spillovers from the Venezuelan Crisis Migration Flows and Their Impact on Latin America and the Caribbean. Departmental Paper DP/2022/019, Western Hemisphere Department, IMF, Washington, D.C.

### Regional and case-study focus areas reflected in the references
- Africa and Sub‑Saharan Africa: multiple contributions on resources, conflict, commodity prices, and demography (e.g., Adhvaryua et al. (2021); Brückner and Ciccone (2010); McGuirk and Burke (2020)).  
- Middle East and North Africa: economic impact and refugee crisis analyses (e.g., Rother et al. (2016)).  
- Latin America: commodity shocks and conflict (e.g., Dube and Vargas (2013); Bazzi et al. (2022)).  
- Country- and event-specific histories and analyses: Sudan (Moro (2006); Rone (2003)); Ivory Coast (Woods (2003)); Russian Revolution (Figes (1996)).

### Interdisciplinary approaches and theoretical foundations
- Political economy and institutional origins: Acemoglu, D., S. Johnson, and A.J. Robinson. (2001). The Colonial Origins of Comparative Development: An Empirical Investigation. American Economic Review, 91 (5):1369–1401.  
- Rational-choice and bargaining approaches to conflict: Hirshleifer, J. (1991). The Paradox of Power. Economics&Politics, 3(3):177–200; Powell, R. (2002). Bargaining Theory and International Conflict. Annual Review of Political Science, 5:1–30.  
- Labor, recruitment, and rebel dynamics: Weinstein, J.M. (2005). Resources and the Information Problem in Rebel recruitment. Journal of Conflict Resolution, 49(4):598–624.  
- Methodological critiques and inference: Ciccone, A. (2011). Economic Shocks and Civil Conflict: A Comment. American Economic Journal: Applied Economics, 3 (4): 215–227; Miguel and Satyanath (2011).

_ Macroeconomic Shocks and Conflict Working Paper No. WP/2023/068 _

---


_Source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023068-print-pdf.pdf_