## wpiea2025012-print-pdf

## Source details

**Canonical URL:** [wpiea2025012-print-pdf](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025012-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025012-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025012-print-pdf.pdf.json)

---

### Introduction — motivation and mechanisms
- Conflicts and conflict-related deaths have been on the rise globally (USIP 2023).
- Conflicts impose large macroeconomic costs and have repercussions on development outcomes including in health and education.
- Paper objective: test whether conflicts predict higher gender inequalities in both outcomes and opportunities using global data.
- Mechanisms by which conflicts affect women:
  - Immediate impacts: death tolls and displacement affecting workforce and human capital.
  - Medium/long-run impacts: destruction of physical infrastructure, erosion of institutions, disruptions in healthcare and education.
  - Gender-specific harms: human rights abuses, trafficking, gender-based violence, impaired access to reproductive health services and maternal care, reduced educational and employment opportunities.
  - Social-role changes: traditional gender roles may become more pronounced or shift, impeding employment and financial security.
- Implications for growth and development:
  - Reversal of progress toward gender equality can hinder economic growth and development.
  - More gender-equal laws increase female labor force participation and facilitate cross-country income convergence (cited literature).

### Data, sample, and measurement
- Main sample: 158 advanced, emerging market and developing economies over 1990–2019 (restricted by data availability); 4,220 country-year observations.
- Conflict data: Major Episodes of Political Violence (MEPV) database (events defined as human toll exceeding 500 deaths; includes interstate and intrastate violence).
  - Baseline coding: Dummy = 1 for all years of conflicts (interstate + intrastate).
  - Alternative codings: separate dummies for interstate/intrastate; onset-year dummy; intensity coding using MEPV scores (0/1/2 based on median).
- Gender inequality measures:
  - Outcomes: Gender Inequality Index (GII) from UNDP, range 0–100, higher values = greater gender inequality (data from 1990).
  - Granular outcomes: female enrollment (primary/secondary/tertiary), percent seats in parliament held by women, female labor force participation (ages 15+), adolescent fertility (births per 1000 women aged 15–19), maternal mortality (deaths per million live births).
  - Opportunities/legal discrimination: World Bank Women, Business and the Law (WBL) index (1970–2018 sample available), overall and eight subindices (mobility; workplace; pay; marriage; parenthood; entrepreneurship; assets; pension). WBL scaled 0–100, higher = greater gender equality in legislation.
- Other controls:
  - Institutional quality: Freedom House civil liberties index (transformed: 8 minus index).
  - Macroeconomic controls: real GDP per capita (log, constant 2015 US$), trade (% of GDP), FDI (% of GDP), credit to private sector (% of GDP), inflation (GDP deflator, percent), agricultural employment (% of total employment). Additional variables for weighted regressions: GDP (constant 2015 US$), population (log).

### Stylized facts and sample statistics
- Conflict observations: 772 country-year observations with conflicts; average annual probability ≈ 18 percent; intrastate = 727; interstate = 70; both types = 25. Sub-Saharan Africa: 247 conflict events.
- Evolution of GII (1990–2019, sample):
  - Simple average of GII decreased from 50 to 35.
  - Simple linear extrapolation implies it may take globally almost 70 more years to reach full gender equality in outcomes (GII = 0) if pace from 1990–2019 continues.
  - 25th percentile of GII distribution = 17 in 2019; 75th percentile = 53 in 2019.
  - Distribution shifts: low GII (<25) 18 percent in 1990 → 35 percent in 2019; medium-low (25–50) 22 percent → 37 percent; medium-high (50–75) 56 percent → 27 percent.
- Macroeconomic relevance:
  - Local projections: GDP per capita decreases about 8 percent on average following conflicts; remains about 5.7 percentage points lower compared to the initial value 10 years after conflict.
  - Cross-country growth association with GII (sample averages): countries below median GII: growth ≈ 2.4 percent per annum; countries above median GII: growth ≈ 1.5 percent per annum.

### Empirical strategy
- Baseline panel specification:
  - Gender inequality_{c,t} = β1 Conflict_{c,t−1} + θ_c + θ_t + ε_{c,t}
  - Extended specification includes lagged controls: Gender inequality_{c,t} = β1 Conflict_{c,t−1} + β2 X_{c,t−1} + θ_c + θ_t + ε_{c,t}
- Identification: country fixed effects (θ_c) and year fixed effects (θ_t); standard errors clustered at country level (unless noted).
- Inference caveat: fixed-effects and controls mitigate omitted variable concerns but do not definitively establish causal inference.

### Main results — composite GII
- Baseline result (Table 1, column 1):
  - Conflict coefficient = 1.34** (0.62).
  - Interpretation: GII becomes 1.3 points higher following years of conflicts.
  - Economic significance:
    - More than two times the average annual change of GII in the sample (average annual change = -0.6 points).
    - Corresponds to 80 percent of one standard deviation of the annual changes across the sample (annual changes SD = 1.6 points).
    - Coefficient > one standard deviation of annual changes for about three quarters of countries (114 out of 158).
- Selected conflict coefficient estimates across specifications: 1.34** (0.62); 1.42** (0.62); 1.30** (0.64); 1.32** (0.61); 1.24** (0.61); 1.40** (0.61); 1.37** (0.62); 1.41** (0.69).
- Control variables (institutional quality, GDP per capita, trade, FDI, credit, inflation, agriculture share) are generally statistically insignificant in explaining one-year changes in GII in these specifications.
- Sample sizes and R-squared across Table 1 columns: observations = 4220, 4188, 4144, 3921, 3850, 2988, 2980, 2880; R-squared = 0.97, 0.97, 0.97, 0.97, 0.97, 0.98, 0.98, 0.98.

### Robustness and placebo evidence
- Robustness checks (Table 2, Panel A — alternative codings and samples):
  - Onset-year dummy: Conflict = 1.07** (0.52).
  - Intensity coding (1/2): Conflict = 0.93** (0.39).
  - Controls for number of past events: Conflict = 1.35** (0.62); coefficient on number of past events ≈ -0.05 (p-value = 0.54).
  - Continuing sample (102 countries): Conflict = 1.31** (0.63).
  - Subsample with at least one event: Conflict = 1.26** (0.62).
  - Excluding countries with ≥ 20 events: Conflict = 1.67*** (0.63).
  - Countries with ≥ 3 events: Conflict = 1.34** (0.64).
  - Excluding poorest countries: Conflict = 1.39** (0.66).
  - Population-weighted regressions: Conflict = 1.31** (0.60).
- Robustness checks (Table 2, Panel B — estimators and fixed effects):
  - GII winsorized at 1st/99th percentiles: Conflict = 1.35** (0.62).
  - Replace year fixed effects with common year trend: Conflict = 1.35** (0.61).
  - Replace country fixed effects with region fixed effects: Conflict = 6.11*** (1.23) (note R-squared falls to 0.80).
  - Replace year fixed effects with decade fixed effects: Conflict = 1.49** (0.60).
  - Heteroskedasticity-robust standard errors: Conflict = 1.34*** (0.23).
- Placebo tests:
  - Procedure: randomly assign placebo conflict dummy maintaining country-specific overall probability; 1000 placebo draws; distribution of t-statistics reported in Figure A.1.
  - Reported summary: 1.4 percent of the placebo tests had a larger t-statistic than the actual result (text truncated in source but this 1.4 percent figure is reported).

### Timing and dynamics around onset (Figure 4)
- Extended event study including lags and leads (r = −4,...,4):
  - Pre-event years do not predict significantly higher GII (confidence intervals encompass zero).
  - Gender inequality becomes higher in years following onset and remains elevated upwards of 4 years after onset.
  - Interpretation: mitigates reverse-causality concerns that higher GII precedes conflicts; supports interpretation of post-event increases.

### Heterogeneity by event type and region
- Types of events (Table 3):
  - Interstate: 1.63** (0.71); 1.62** (0.72); 1.66** (0.72) across columns.
  - Intrastate: 1.14* (0.65).
  - Civil: 2.07** (0.85).
  - Ethnic: -0.30 (0.83) (no strong evidence; note categorization caveats).
- Distance to frontier and dispersion (Table 4 and Figure 5):
  - Countries fall behind gender-equality frontiers (10th/5th/1st percentiles and minimum) by roughly 1.33–1.55 points after conflicts (Conflict coefficients reported: 1.41** (0.61); 1.55** (0.60); 1.42** (0.61); 1.35** (0.61); 1.33** (0.63) for continuing sample).
  - Cross-country dispersion: standard deviation of GII each year positively correlated with number of events in previous year; correlation = 0.66.
- Regional results (Table 5):
  - World Bank regions (Panel A, Conflict coefficients):
    - East Asia and Pacific (EAP): 1.14** (0.55).
    - Europe and Central Asia (ECA): -0.21 (0.69).
    - High income OECD (HI OECD): 0.43 (0.40).
    - Latin America and Caribbean (LAC): -0.33 (0.50).
    - Middle East and North Africa (MENA): 4.44*** (0.97).
    - South Asia (SA): -2.50*** (0.85).
    - Sub-Saharan Africa (SSA): 1.24*** (0.27).
  - MEPV regions (Panel B, Conflict coefficients):
    - Africa: 1.11*** (0.28).
    - America: -0.47 (0.51).
    - Asia: 2.16*** (0.56).
    - Europe: 1.08 (0.67).
  - Interpretation: adverse associations pronounced in EAP, MENA, SSA (World Bank regions) and in Africa and Asia (MEPV classification).

### Granular outcomes (Table 6)
- Conflict coefficients by specific gender outcomes (selected estimates and interpretation):
  - Primary female enrollment: -5.63*** (1.99).
    - One standard deviation of annual changes in female primary enrollment rate = 3.8 percentage points.
  - Secondary female enrollment: -0.79 (1.74).
  - Tertiary female enrollment: 2.45 (1.87).
  - Female labor force participation: 0.33 (0.70).
  - Share of women in parliament: -1.46* (0.87).
    - Annual changes SD = 2.3 percentage points.
  - Adolescent fertility rate: 2.68 (1.65).
  - Maternal mortality ratio: 6.55** (2.68).
    - One standard deviation of annual changes in maternal mortality = 1.6.
- Interpretation:
  - Primary female enrollment falls by 5.6 percentage points post-conflict; maternal mortality increases by 6.55 per million live births; share of women in parliament declines by 1.46 percentage points.
  - No strong evidence of immediate changes in secondary/tertiary enrollment or female labor force participation in the year following conflicts, though primary enrollment declines imply potential medium-term adverse effects.

### Gender inequality in opportunities — legal rights (Table 7)
- Sample: 163 countries, 1970–2018, 7,404 observations.
- Conflict coefficients on WBL index and subindices (negative = decline in legal equality; WBL scaled 0–100, higher = greater gender equality):
  - WBL overall: -2.45*** (0.78).
  - Mobility: -1.59** (0.73).
  - Workplace: -2.84 (1.96).
  - Pay: -1.88 (1.33).
  - Marriage: -2.62* (1.46).
  - Parenthood: -3.46** (1.43).
  - Entrepreneurship: -4.13*** (1.53).
  - Assets: -1.00 (1.26).
  - Pension: -2.05 (1.66).
- Interpretation:
  - Conflicts predict a 2.5 point decline in WBL overall (one standard deviation of annual changes in WBL across sample = 2.4 points).
  - Negative effects concentrated in mobility, marriage, parenthood, entrepreneurship subindices, indicating setbacks in legal gender equality following conflicts.

### Summary of key quantitative findings
- Conflict predicts higher GII by roughly 1.3–1.6 points depending on specification and sample.
- Primary female enrollment falls by 5.63 percentage points following conflicts.
- Share of women in parliament declines by 1.46 percentage points following conflicts.
- Maternal mortality increases by 6.55 (per million live births) following conflicts.
- Conflicts reduce WBL overall by -2.45 points and notably harm mobility, parenthood, and entrepreneurship subindices.
- Countries fall behind gender-equality frontiers by roughly 1.33–1.55 points after conflicts.
- Cross-country dispersion in GII is positively correlated with number of conflict episodes (correlation = 0.66).

### Conclusion and policy implications
- Conflicts are associated with a durable drag on development and predict higher gender inequality in outcomes and opportunities.
- Key implications:
  - Both interstate and intrastate conflicts predict higher gender inequality.
  - Post-conflict setbacks in education, political representation, maternal health, and legal rights likely contribute to persistent development costs.
- Policy-relevant recommendations (implied by empirical findings):
  - Immediate action in conflict-affected states to address gender inequalities in outcomes and opportunities is required to support sustainable and inclusive growth.
  - Priority areas include addressing educational disruptions (notably primary female enrollment), restoring health services (maternal care), protecting women’s political participation, and restoring legal gender rights across mobility, marriage, parenthood, and entrepreneurship.

*Source: IMF Working Paper — wpiea2025012-print-pdf (1. Introduction; 2. Data; 5. Results; 6. Conclusion sections of the PDF)*

### 1. Introduction ........................................................................................................

### 1. Introduction

### Context and motivation
- Conflicts and conflict-related deaths have been on the rise globally (USIP 2023).
- Conflicts impose large macroeconomic costs and have repercussions on development outcomes including in health and education.
- Gender equality is key for long-term sustained economic growth; existing literature has focused more on the role of gender disparities in the occurrence of conflicts than on how conflicts affect gender inequality.
- This paper argues that conflicts likely affect women more than men and tests this using global data to show that conflicts predict higher gender inequalities in both outcomes and opportunities.

### Mechanisms by which conflicts affect women
- Immediate impacts: death tolls and displacement affecting workforce and human capital.
- Medium/long-run impacts: destruction of physical infrastructure, erosion of institutions, disruptions in healthcare and education.
- Gender-specific harms: human rights abuses, trafficking, gender-based violence, impaired access to reproductive health services and maternal care, reduced educational and employment opportunities.
- Social-role changes: traditional gender roles may become more pronounced or shift (e.g., women as caretakers), impeding employment and financial security (Sommer et al. 2018, Dincecco et al. 2024).
- UN reporting indicates women’s empowerment may be negatively impacted during conflicts, limiting access to employment and financial resources.

### Implications for growth and development
- Reversal of progress toward gender equality due to wars and instability can hinder economic growth and development (Boserup 1970; Bertay et al. 2020, 2024; Sever 2023, 2022; Galor and Weil 1996; Lagerlöf 2003).
- Evidence that more gender-equal laws increase female labor force participation and facilitate cross-country income convergence (Sever 2023, Sever 2022).
- Gender inequality is linked to negative growth outcomes; conflicts may exacerbate these channels.

### Data, sample, and measures used
- Main sample: 158 advanced, emerging market and developing economies over the period of 1990 and 2019, restricted by data availability.
- Outcome measure: gender inequality index (GII) from UNDP as a composite indicator of gender disparities in health, education, political representation and labor market.
- Conflict data: Major Episodes of Political Violence (MEPV) database from the Center of Systemic Peace.
- Legal/opportunity measure: World Bank’s Women, Business and the Law (WBL) database measuring laws and regulations since 1970 across eight areas of women’s economic rights (e.g., freedom of movement, rights to employment and equal remuneration, protection from harassment, paid maternity leave benefits, access to credit, pensions, and assets).

### Stylized facts documented
- There has been steady progress toward gender equality globally since the 1990s, but gender inequalities remain prevalent and cross-country gaps do not seem to narrow over time.
- Suggestive evidence that greater gender inequality in outcomes is associated with lower economic growth on average.
- Conflicts are widespread across regions and years in the sample.
- Conflicts are associated with a long-lasting decline in GDP per capita on average, upwards of 10 years following those events.

### Baseline empirical findings
- Conflicts predict higher gender inequality in outcomes (using GII).
- Identification approach includes country and year fixed effects to account for time-invariant country-specific factors and common yearly shocks.
- The finding remains similar when isolating variation in gender inequality from a large set of time-varying institutional and macroeconomic variables.
- No strong evidence that higher gender inequality precedes the onset of conflicts, mitigating reverse causality concerns.
- The negative association between conflicts and gender equality survives a large set of robustness checks and placebo tests, supporting a systematic relationship between conflicts and gender inequality.

### Cross-country dispersion and convergence
- Conflicts tend to exacerbate cross-country gaps in gender equality:
  - Conflicts predict a higher distance between a country and the global frontier regarding gender equality.
  - Conflicts weaken sigma-convergence across countries—cross-country dispersion in gender equality is greater following years with more frequent conflicts.
- These patterns are especially relevant for regions with relatively greater gender inequalities and more frequent conflicts, such as Sub-Saharan Africa.

### Heterogeneity: types of conflict and specific outcome components
- Both interstate and intrastate conflicts are associated with higher gender inequality.
- Using granular outcome measures (instead of the composite index):
  - Female primary school enrollment rate and share of seats in parliament held by women tend to decrease after conflicts.
  - Maternal mortality rate increases after conflicts.
  - These declines suggest impediments to human capital accumulation following conflicts.

### Conflicts and gender inequality in opportunities (legal discrimination)
- Conflicts are associated with less economic rights for women as measured by the WBL database, aggravating unequal treatment under the law.
- Effects are particularly pronounced in laws related to freedom of mobility, rights in marriage and parenthood, and entrepreneurship.
- These results imply conflicts impede progress toward leveling the legal playing field between women and men.

### Contribution to literatures
- Adds to literature linking conflicts and gender inequality by documenting that conflicts predict higher gender inequality in outcomes and opportunities using cross-country composite measures.
- Augments literature on economic costs of conflicts by identifying gender inequality as a channel through which conflicts may produce long-lasting negative effects on economic performance (Collier 1999; Hoeffler and Reynal-Querol 2003; Cerra and Saxena 2008; Gates et al. 2012; Fang et al. 2020; Novta and Pugacheva 2021; Chupilkin and Koczan 2022; Rodrik 1999; Sever 2024; Ghobarah et al. 2003; Plümper and Neumayer 2006).
- Contributes to literature on drivers of gender inequality by showing conflicts are a significant predictor of gender inequality even after controlling for slow-moving variables and other determinants (Cooper 2002; Fernandez 2007; Branisa et al. 2013; Alesina et al. 2013; Cooray and Potrafke 2011; Ferrant and Nowacka 2015; Jayachandran 2015; Klasen 2017; Nguyen 2022; Goldin 2023).

### Policy-relevant implications
- Conflicts’ tendencies to increase gender inequality provide a potential channel for their long-lasting negative effects on economic growth and development.
- Given the COVID-19 pandemic likely exacerbated existing gender gaps (e.g., Alon et al. 2020), ongoing political and security challenges heighten the importance of addressing conflict-related gender disparities for sustainable and inclusive post-pandemic growth.

*Source: IMF Working Paper — 1. Introduction section of wpiea2025012-print-pdf*

### 2. Data

### 2. Data

### Episodes of Conflicts
- Data source: Major Episodes of Political Violence (MEPV) database from Center for Systemic Peace (available until 2018).
- Definition: Conflicts identified by systematic and sustained use of violence by organized groups where human toll exceeds 500 deaths; includes interstate and intrastate violence and war (civil and ethnic subgroups).
- Main coding:
  - Dummy variable = 1 for all years of conflicts (interstate and intrastate) in baseline.
  - Alternative codings tested:
    - Separate dummies for interstate and intrastate conflicts.
    - Split intrastate into civil and ethnic violence (noting the distinction is not always clear).
    - Dummy = 1 only for onset year of conflicts.
  - Conflict “size” accounted for in a separate test using MEPV scores (1 to 10):
    - Variable = 1 when score is below the sample median.
    - Variable = 2 when score is above the sample median.
    - Variable = 0 when no conflict.
  - Rationale: median-based classification preferred over exact scores due to comparability concerns across countries and years.

### Gender Inequality
#### 2.2.2. Gender Inequality in Outcomes
- Baseline measure: composite index in gender inequality (GII) from United Nations Development Programme (UNDP).
  - Data available starting from 1990.
  - Index range: 0 to 100, higher values = greater gender inequality.
- Additional outcome variables used (data sources specified):
  - Female school enrollment rates in primary, secondary and tertiary levels (gross rates) — World Bank’s World Development Indicators.
  - Percent of seats in parliament held by women — UNDP.
  - Female labor force participation rate (ages 15 and older) — UNDP.
  - Adolescent fertility rate (births per 1000 women aged 15-19) — UNDP.
  - Maternal mortality rate (deaths per million live births) — UNDP.
- Empirical note: Table A.3 shows granular gender-based outcomes are positively associated with GII; gender inequality in laws is correlated with gender inequality in outcomes.

#### 2.2.2. Gender Inequality in Opportunities (Legislation)
- Data source: World Bank’s Women, Business and the Law (WBL) database (since 1970).
- Scope: eight indicator categories covering women’s work life from entry to retirement:
  - (i) mobility, (ii) workplace, (iii) pay, (iv) marriage, (v) parenthood, (vi) entrepreneurship, (vii) ownership of assets, (viii) pension.
- Scoring:
  - Each indicator based on 4 or 5 binary (Yes/No) questions.
  - A legal limitation on women assigns 0 for that question.
  - For indicators with 4 questions, each “Yes” adds 25 points; with 5 questions, each “Yes” adds 20 points.
  - Indicator-level score = unweighted sum of question scores within that indicator.
  - Overall WBL index = unweighted average of eight indicator scores for each country-year.
  - WBL index scaled between 0 and 100, higher values = greater gender equality in legislation.
  - Interpretation examples:
    - WBL = 25 means women, on average, have a quarter of the economic rights that men have.
    - WBL = 100 indicates total equality between women and men in legislation.

### Other Variables
- Institutional quality proxy: index on civil liberties from Freedom House.
  - Original range: 1 to 7, higher values = lower degree of civil liberties.
  - Transformed in regressions: 8 minus the index, so higher values indicate a better institutional environment.
- Macroeconomic control variables (World Bank World Development Indicators):
  - Real GDP per capita (used in logarithm, constant in 2015 US dollars). Note: nominal GDP per capita used for a few observations where real data missing (does not affect results).
  - Trade (as percent of GDP).
  - FDI (net inflows, as percent of GDP).
  - Credit to the private sector (as percent of GDP).
  - Inflation (based on GDP deflator, percent).
  - Employment in agriculture (modeled estimates, as percent of total employment).
- Additional variables for weighted regressions:
  - GDP (constant in 2015 US dollars).
  - Population (used in logarithm).
- Empirical note: Table A.3 shows gender inequalities in outcomes are more prevalent in countries that are less developed.

### Key sample statistics and distributions (from the sample described)
- Main sample: 158 countries (advanced, emerging market and developing economies) over the period of 1990-2018 (restricted by MEPV and UNDP availability).
- Observations: 4,220 country-year observations in the sample.
- Conflict observations:
  - 772 country-year observations with conflicts.
  - Average annual probability of conflicts: about 18 percent.
  - Intrastate conflict observations: 727.
  - Interstate conflict observations: 70.
  - Observations with both types: 25.
- Regional note: sub-Saharan Africa leads with 247 conflict events (also the region with highest number of observations in sample).

### Evolution of Gender Inequality (stylized facts)
- Population-weighted/simple averages and percentiles:
  - Simple average of GII in the sample decreased from 50 to 35 from 1990 to 2019.
  - Simple linear extrapolation implies it may take globally almost 70 more years to reach full gender equality in outcomes (GII = 0) if pace from 1990-2019 continues.
  - 25th percentile of GII distribution reached 17 in 2019.
  - 75th percentile of GII distribution was 53 in 2019.
- Distributional shifts (percentage of countries by GII category):
  - Categories: GII below 25 (low); between 25 and 50 (medium-low); between 50 and 75 (medium-high); above 75 (high).
  - Countries with low GII: 18 percent in 1990 → 35 percent in 2019.
  - Medium-low GII: 22 percent in 1990 → 37 percent in 2019.
  - Medium-high GII: 56 percent in 1990 → 27 percent in 2019.

### Macroeconomic relevance of conflicts and gender inequality (stylized facts)
- Local projections (Jorda 2005) results:
  - GDP per capita decreases about 8 percent on average following conflicts.
  - GDP per capita remains about 5.7 percentage points lower compared to the initial value 10 years after the conflict (negative association statistically significant for upwards of 10 years).
- Cross-country growth association with GII (sample averages):
  - Countries below the sample median GII (low gender inequality): growth rate of real GDP per capita ≈ 2.4 percent per annum on average.
  - Countries above the sample median GII (higher gender inequality): growth rate of real GDP per capita ≈ 1.5 percent per annum on average.
- Interpretation: patterns suggest higher gender inequality in the aftermath of conflicts can be a potential channel for persistent declines in GDP per capita after conflicts.

### Methodology (empirical specifications)
- Baseline panel regression with fixed effects (equation (1)):
  - Gender inequality_{c,t} = β1 Conflict_{c,t−1} + θ_c + θ_t + ε_{c,t}
  - Notation: c = country, t = year.
  - Gender inequality proxy: GII (UNDP), higher values = greater gender inequality in outcomes.
  - Conflict_{c,t−1}: dummy = 1 during years of conflicts (interstate + intrastate), 0 otherwise.
  - Expectation: if conflicts predict higher gender inequality, β1 > 0.
  - θ_c = country fixed effects; θ_t = year fixed effects.
- Extended specification with controls (equation (2)):
  - Gender inequality_{c,t} = β1 Conflict_{c,t−1} + β2 X_{c,t−1} + θ_c + θ_t + ε_{c,t}
  - X_{c,t−1} = set of control variables (institutional and macroeconomic environment variables described above).
- Inference:
  - Standard errors clustered at the country level in estimations unless otherwise noted.
  - Fixed effects absorb time-invariant country factors and common annual shocks; inclusion of controls intended to mitigate omitted variable concerns though causality is not definitively claimed.

*Source: wpiea2025012-print-pdf - 2. Data*

### 5. Results

### 5. Results

### 5.1. Main Results
- Dependent variable: composite index on gender inequality (GII) as adopted from UNDP.
- Key explanatory variable: Conflict — dummy indicating country-year observations with inter- or intra-state wars or violence.
- Main finding (Table 1, column 1):
  - Conflict coefficient = 1.34** (standard error = 0.62).
  - Interpretation: gender inequality index becomes 1.3 points higher following the years of conflicts.
  - Economic significance:
    - This is more than two times the average annual change of GII in the sample (which is -0.6 points).
    - It corresponds to 80 percent of one standard deviation of the annual changes across the sample (which is 1.6 points).
    - The coefficient is larger than one standard deviation of the annual changes in GII for about three quarters of the countries (i.e., for 114 countries out of 158 in the sample).
- Controls and specification:
  - Country fixed effects and year fixed effects included in all reported specifications.
  - Additional control variables tested (columns 2–8): institutional quality (index on civil liberties), logarithm of real GDP per capita, trade as percent of GDP, FDI as percent of GDP, credit to the private sector as percent of GDP, inflation (deflator-based), agricultural employment as share of total employment.
  - Coefficient estimates on these control variables are statistically insignificant across tests.
  - Sample size and fit (Table 1):
    - Observations by column: 4220, 4188, 4144, 3921, 3850, 2988, 2980, 2880.
    - R-squared by column: 0.97, 0.97, 0.97, 0.97, 0.97, 0.98, 0.98, 0.98.
  - Note: sample shrinks by one-third in the last column relative to column 1; therefore control variables are dropped for the rest of the analysis.

- Selected coefficients from Table 1 (exact values):
  - Conflict: 1.34** (0.62); 1.42** (0.62); 1.30** (0.64); 1.32** (0.61); 1.24** (0.61); 1.40** (0.61); 1.37** (0.62); 1.41** (0.69).
  - Institutional quality: 0.20 (0.28); 0.30 (0.25); 0.10 (0.22); 0.07 (0.22); 0.12 (0.24); 0.11 (0.24); 0.00 (0.27).
  - GDP per capita: -0.11 (2.08); -1.06 (1.31); -1.00 (1.36); -1.82 (1.29); -1.55 (1.47); -1.42 (1.42).
  - Trade: 0.00 (0.01); 0.00 (0.01); 0.01 (0.01); 0.01 (0.01); 0.01 (0.01).
  - FDI: -0.01 (0.01); 0.00 (0.01); 0.00 (0.01); 0.00 (0.01).
  - Financial development: 0.00 (0.01); 0.00 (0.01); 0.00 (0.01).
  - Inflation: 0.00 (0.00); 0.00 (0.00).
  - Agriculture share: -0.02 (0.06).

- Inference:
  - The positive and statistically significant association between conflicts and higher GII is robust to inclusion of many annual controls.
  - Insignificant coefficients on controls suggest those variables are not associated with pronounced one-year changes in GII in this sample.
  - Country fixed effects are important; the analysis focuses on annual changes while accounting for time-invariant country factors.

### 5.2. Robustness
- Overview: a large set of robustness checks reported in Table 2, Panels A and B.

- Panel A — alternative conflict codings and sample choices (Table 2, Panel A):
  - Column 1: dummy = 1 only for onset year of events.
    - Conflict = 1.07** (0.52).
  - Column 2: intensity coding (1 for milder events, 2 otherwise).
    - Conflict = 0.93** (0.39).
  - Column 3: controls for number of years with events in each country (over the period) until year t-1.
    - Conflict = 1.35** (0.62).
    - Note: coefficient on number of past events ≈ -0.05, p-value = 0.54 (statistically insignificant).
  - Column 4: continuing sample (102 countries with data for all years).
    - Conflict = 1.31** (0.63).
  - Column 5: subsample of countries with at least one event during the period.
    - Conflict = 1.26** (0.62).
  - Column 6: excludes countries with 20 or more events.
    - Conflict = 1.67*** (0.63).
  - Column 7: countries with 3 or more events during the period.
    - Conflict = 1.34** (0.64).
  - Column 8: excludes poorest countries (based on 5th percentile of average real GDP per capita).
    - Conflict = 1.39** (0.66).
  - Column 9: weighted regressions with weights = average population for each country.
    - Conflict = 1.31** (0.60).
  - Observations and fit (Panel A):
    - Observations by column: 4220, 4220, 4220, 3060, 1982, 3922, 1634, 4018, 4220.
    - R-squared by column: 0.97, 0.97, 0.97, 0.97, 0.97, 0.94, 0.98, 0.97, 0.97.

- Panel B — alternative estimators, fixed effects, and outlier treatment (Table 2, Panel B):
  - Column 1: GII winsorized at 1st and 99th percentiles.
    - Conflict = 1.35** (0.62).
  - Column 2: replace year fixed effects with a common year trend.
    - Conflict = 1.35** (0.61).
  - Column 3: replace country fixed effects with region fixed effects.
    - Conflict = 6.11*** (1.23).
    - R-squared = 0.80 (country fixed effects not included).
  - Column 4: replace year fixed effects with decade fixed effects.
    - Conflict = 1.49** (0.60).
  - Column 5: use heteroskedasticity-robust standard errors instead of clustering.
    - Conflict = 1.34*** (0.23).
  - Observations and fit (Panel B):
    - Observations for all columns: 4220.
    - R-squared by column: 0.97, 0.97, 0.80, 0.96, 0.97.

- Additional robustness and placebo testing:
  - Results robust to winsorizing GII at 5th and 95th percentiles (unreported).
  - Results remain similar when replacing or dropping fixed effects; however, dropping country fixed effects reduces R-squared markedly (to about 0.1), indicating the importance of country-level time-invariant factors.
  - Standard errors results:
    - Results similar when using two-way clustering (country- and year-level).
  - Placebo (randomization) tests:
    - Procedure: calculate overall probability of events for each country, randomly assign placebo conflict dummy maintaining that probability, create 1000 placebo event sets, run tests.
    - Figure A.1 in the Appendix shows distribution of t-statistics across 1000 tests; only about (text truncated in source).

*Source: wpiea2025012-print-pdf - 5. Results*

### 1.4 percent of the tests with the placebo dummy variable has a larger t-statistic than the one that we

### wpiea2025012-print-pdf - 1.4 percent of the tests with the placebo dummy variable has a larger t-statistic than the one that we

### 5.3. Gender Inequality around the Onset of Conflicts
- Extended estimation includes the four years of forward values of the dummy for the onset of conflicts, the dummy at year t, and 4 lags of it.
- Findings (Figure 4):
  - The years running up to conflicts do not predict significantly higher gender inequality (confidence interval encompasses zero).
  - Gender inequality becomes higher in the years following such events.
  - Gender inequality remains high upwards of 4 years in the aftermath of the onset of conflicts.
- Interpretation: Mitigates concern that higher pre-existing gender inequality drives entry into conflicts; supports causal interpretation of post-event increases.

### 5.4. Types of Events
- Decomposition by event type (Table 3):
  - Interstate coefficient (columns 1–3): 1.63** (0.71); 1.62** (0.72); 1.66** (0.72).
  - Intrastate (column 2): 1.14* (0.65).
  - Civil (column 3): 2.07** (0.85).
  - Ethnic (column 3): -0.30 (0.83).
- Controls and fit:
  - Country F.E.: Yes (all columns).
  - Year F.E.: Yes (all columns).
  - Observations: 4220 (all columns).
  - R-squared: 0.97 (all columns).
- Interpretation:
  - Both interstate and intrastate episodes are associated with higher gender inequality.
  - Stronger roles found for interstate and civil violence; little evidence for ethnic-based events (caution due to categorization issues).

### 5.5. Distance to the Frontier and Dispersion in Gender Equality
- Dependent variable: country GII minus frontier GII; frontiers defined by the 10th, 5th, 1st percentiles and the minimum each year. Decade fixed effects used.
- Table 4 coefficient estimates for Conflict:
  - Column 1: 1.41** (0.61).
  - Column 2: 1.55** (0.60).
  - Column 3: 1.42** (0.61).
  - Column 4: 1.35** (0.61).
  - Column 5 (continuing sample, 10th percentile frontier): 1.33** (0.63).
- Controls and fit:
  - Country F.E.: Yes (all).
  - Decade F.E.: Yes (all).
  - Observations: 4220, 4220, 4220, 4220, 3060.
  - R-squared: 0.97, 0.97, 0.97, 0.97, 0.97.
- Interpretation:
  - Countries fall behind peers at the frontier in gender equality by roughly 1.33–1.55 points in the aftermath of conflicts across frontier definitions.
  - Conflicts likely widen cross-country gaps in gender equality.
- Cross-country dispersion (Figure 5):
  - Standard deviation of GII each year is positively correlated with the number of events in the previous year.
  - Correlation between standard deviation of GII and number of episodes: 0.66.
  - Suggestive evidence that conflicts tend to widen gaps in gender equality across countries, with implications for regions like Sub-Saharan Africa.

### 5.6. Results in Different Regions
- Panel A (World Bank regions, Table 5) — Conflict coefficients by region:
  - East Asia and Pacific (EAP, column 1): 1.14** (0.55).
  - Europe and Central Asia (ECA, column 2): -0.21 (0.69).
  - High income OECD (HI OECD, column 3): 0.43 (0.40).
  - Latin America and Caribbean (LAC, column 4): -0.33 (0.50).
  - Middle East and North Africa (MENA, column 5): 4.44*** (0.97).
  - South Asia (SA, column 6): -2.50*** (0.85).
  - Sub-Saharan Africa (SSA, column 7): 1.24*** (0.27).
- Panel A controls and fit:
  - Country F.E.: Yes (all).
  - Year F.E.: Yes (all).
  - Observations: 371, 548, 973, 639, 414, 175, 1100.
  - R-squared: 0.97, 0.92, 0.93, 0.93, 0.83, 0.95, 0.94.
- Panel B (MEPV regions, Table 5) — Conflict coefficients:
  - Africa (column 1): 1.11*** (0.28).
  - America (column 2): -0.47 (0.51).
  - Asia (column 3): 2.16*** (0.56).
  - Europe (column 4): 1.08 (0.67).
- Panel B controls and fit:
  - Country F.E.: Yes (all).
  - Year F.E.: Yes (all).
  - Observations: 1180, 669, 1103, 1106.
  - R-squared: 0.93, 0.92, 0.92, 0.94.
- Interpretation:
  - Positive association between conflicts and gender inequality is pronounced in East Asia and Pacific, Middle East and North Africa, and Sub-Saharan Africa (Panel A).
  - Using MEPV classification, adverse role is pronounced in Africa and Asia (Panel B).
  - These regions account for a disproportionate share of events in the sample.

### 5.7. Granular Outcomes
- Table 6 — Conflict coefficients by specific gender outcomes:
  - Primary female enrollment (column 1): -5.63*** (1.99).
  - Secondary female enrollment (column 2): -0.79 (1.74).
  - Tertiary female enrollment (column 3): 2.45 (1.87).
  - Female labor force participation (column 4): 0.33 (0.70).
  - Share of women in parliament (column 5): -1.46* (0.87).
  - Adolescent fertility rate (column 6): 2.68 (1.65).
  - Maternal mortality ratio (column 7): 6.55** (2.68).
- Controls and fit:
  - Country F.E. and Year F.E.: Yes (all columns).
  - Observations: 3545, 2903, 2600, 4220, 4220, 4220, 4220.
  - R-squared: 0.69, 0.94, 0.89, 0.96, 0.84, 0.97, 0.93.
- Interpretation and magnitude:
  - Primary female enrollment falls by 5.6 percentage points post-conflict. One standard deviation of annual changes in female primary enrollment rate is 3.8 percentage points.
  - Share of women in parliament decreases by 1.5 percentage points post-conflict (about two-thirds of one standard deviation; annual changes SD = 2.3 percentage points).
  - Maternal mortality increases by 6.6 (per million live births) post-conflict. One standard deviation of annual changes in maternal mortality is 1.6.
  - No strong evidence of changes in secondary/tertiary enrollment or female labor force participation in the year following conflicts; however, declines in primary enrollment imply potential medium-term adverse effects on higher education, adolescent fertility, labor market outcomes, and political representation.

### 5.8. Gender Inequality in Opportunities (Legal Rights)
- Sample: 163 countries, period 1970–2018.
- Table 7 — Conflict coefficients on WBL index and subcategories (higher values indicate higher gender equality; negative coefficient = decline in equality):
  - WBL overall (column 1): -2.45*** (0.78).
  - Mobility (column 2): -1.59** (0.73).
  - Workplace (column 3): -2.84 (1.96).
  - Pay (column 4): -1.88 (1.33).
  - Marriage (column 5): -2.62* (1.46).
  - Parenthood (column 6): -3.46** (1.43).
  - Entrepreneurship (column 7): -4.13*** (1.53).
  - Assets (column 8): -1.00 (1.26).
  - Pension (column 9): -2.05 (1.66).
- Controls and fit:
  - Country F.E.: Yes (all).
  - Year F.E.: Yes (all).
  - Observations: 7404 (all columns).
  - R-squared: 0.89, 0.94, 0.75, 0.80, 0.89, 0.86, 0.67, 0.86, 0.77.
- Interpretation:
  - Conflicts predict a 2.5 point decline in the WBL index overall (one standard deviation of annual changes in WBL across the sample is 2.4 points).
  - Legal setbacks are widespread across mobility, marriage, parenthood, and entrepreneurship sub-indices, indicating that conflicts harm gender equality in opportunities as well as outcomes.

### 6. Conclusion — Key takeaways and policy implications
- Conflicts are associated with a durable drag on development and specifically predict higher gender inequality in outcomes and opportunities.
- Empirical findings:
  - Both interstate and intrastate conflicts predict higher gender inequality.
  - Post-conflict declines include primary female enrollment (-5.63 percentage points), share of women in parliament (-1.46 percentage points), and increases in maternal mortality (+6.55 per million live births).
  - Countries fall behind peers at the gender-equality frontier by roughly 1.33–1.55 points after conflicts.
  - Conflicts reduce gender equality under law (WBL overall: -2.45 points) and across multiple subdomains.
  - Conflicts are correlated with increased cross-country dispersion in GII (correlation 0.66 between GII standard deviation and number of episodes).
- Policy implications:
  - Immediate action is required to address gender inequalities in outcomes and opportunities in conflict-affected states to support sustainable and inclusive growth.
  - Addressing educational disruptions, restoring health services, protecting women’s political participation, and restoring legal gender rights are critical to prevent long-term setbacks in gender equality and development.

*Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025012-print-pdf.pdf*

### References

### wpiea2025012-print-pdf - References

### Major referenced themes
- Scholarly literature on gender inequality, its origins, and economic consequences:
  - Alesina, Alberto, Paola Giuliano, and Nathan Nunn. “On the origins of Gender Roles: Women and the Plough.” The Quarterly Journal of Economics, Vol. 128, Issue 2, (2013): 469-530.
  - Alon, Titan, Matthias Doepke, Jane Olmstead-Rumsey, and Michèle Tertilt. “The impact of COVID-19 on gender equality.” No. w26947. National Bureau of Economic Research, 2020.
  - Boserup, Ester. “Woman’s Role in Economic Development.” New York: St. Martin’s Press (1970).
  - Jayachandran, Seema. “The Roots of Gender Inequality in Developing Countries”. Annual Review of Economics, Vol. 7, (2015): 63-88.
  - Klasen, Stephan. "Gender, institutions, and economic development." The Handbook of Economic Development and Institutions (2017).
  - Sever, Can. "Legal Gender Equality as a Catalyst for Convergence." IMF Working Papers 2022, no. 155. International Monetary Fund (2022).
  - Sever, Can. “Conflicts and Growth: The R&D Channel.” IMF Working Papers 2024, no. 098. International Monetary Fund (2024).
- Literature on armed conflict and development outcomes:
  - Collier, Paul. "On the economic consequences of civil war." Oxford Economic Papers 51, no. 1 (1999): 168-183.
  - Gates, Scott, Håvard Hegre, Håvard Mokleiv Nygård, and Håvard Strand. "Development consequences of armed conflict." World Development 40, no. 9 (2012): 1713-1722.
  - Ghobarah, Hazem, Huth, Paul, and Bruce Russett. “Civil Wars Kill and Maim People -Long after the Shooting Stops.” The American Political Science Review, Vol. 97, No.2, pp. 189-202. (2003).
  - Hoeffler, Anke, and Marta Reynal-Querol. "Measuring the costs of conflict." Washington, DC: World Bank (2003).
  - Novta, Natalija, and Evgenia Pugacheva. "The macroeconomic costs of conflict." Journal of Macroeconomics 68 (2021): 103286.
- Data sources and institutional indices cited:
  - MEPV (Center for Systemic Peace) — Major Episodes of Political Violence (MEPV) Database.
  - UNDP — Gender Inequality Index; Women in parliament; Labor force participation (female); Adolescent fertility; Maternal mortality.
  - WDI — World Development Indicators: Enrollment (female, primary/secondary/tertiary), GDP (constant 2015 US$ (log)), GDP per capita (constant 2015 US$ (log)), Trade, FDI, Credit, Inflation, Agricultural employment, Population.
  - WBL — Women, Business and the Law indices: WBL index; Mobility; Workplace; Pay; Marriage; Parenthood; Entrepreneurship; Assets; Pension.
  - Freedom House — Civil liberties (proxy for institutional quality).
  - World Bank publications: Women, Business and the Law 2023; World Development Indicators 2023.

### Appendix — Data description (Table A.1)
- Episodes of conflicts: Events of interstate and intrastate violence — Source: MEPV
- Gender inequality index (GII): Composite index on gender inequality in outcomes — Source: UNDP
- Enrollment (female, primary): Primary school enrollment rate, percent, gross — Source: WDI
- Enrollment (female, secondary): Secondary school enrollment rate, percent, gross — Source: WDI
- Enrollment (female, tertiary): Tertiary school enrollment rate, percent, gross — Source: WDI
- Women in parliament: Share of seats in parliament held by women — Source: UNDP
- Labor force participation (female): Participation rate, percent, ages 15 and above — Source: UNDP
- Adolescent fertility: Births per 1000 women ages 15-19 — Source: UNDP
- Maternal mortality: Deaths per million live births — Source: UNDP
- WBL index: Gender equality in legislation, overall index — Source: WBL
- Mobility: Gender equality in legislation, women’s mobility — Source: WBL
- Workplace: Gender equality in legislation, protection in workplace — Source: WBL
- Pay: Gender equality in legislation, equal pay — Source: WBL
- Marriage: Gender equality in legislation, equal rights as spouse — Source: WBL
- Parenthood: Gender equality in legislation, equal rights as parents — Source: WBL
- Entrepreneurship: Gender equality in legislation, rights to start/run business — Source: WBL
- Assets: Gender equality in legislation, rights for asset ownership — Source: WBL
- Pension: Gender equality in legislation, equal treatment in pension — Source: WBL
- Civil liberties: Proxy for institutional quality — Source: Freedom House
- GDP: Constant in 2015 US dollars (log) — Source: WDI
- GDP per capita: Constant in 2015 US dollars (log) — Source: WDI
- Trade: As percent of GDP — Source: WDI
- FDI: Net inflows, as percent of GDP — Source: WDI
- Credit: Credit to the private sector, as percent of GDP — Source: WDI
- Inflation: Deflator-based (percent) — Source: WDI
- Agricultural employment: As percent of total employment — Source: WDI
- Population: Total population (log) — Source: WDI

### Summary statistics (Table A.2)
- Conflicts (dummy variable): Mean 0.18; Median 0; Standard dev. 0.39; Observations 4220
- Gender inequality index (GII): Mean 41.99; Median 45.10; Standard dev. 20.17; Observations 4220
- Enrollment (female, primary): Mean 99.13; Median 101.04; Standard dev. 17.17; Observations 3545
- Enrollment (female, secondary): Mean 77.29; Median 87.31; Standard dev. 33.92; Observations 2903
- Enrollment (female, tertiary): Mean 39.13; Median 33.93; Standard dev. 31.00; Observations 2600
- Women in parliament: Mean 16.68; Median 14.45; Standard dev. 10.74; Observations 4220
- Labor force participation (female): Mean 50.91; Median 50.55; Standard dev. 16.07; Observations 4220
- Adolescent fertility: Mean 60.29; Median 49.50; Standard dev. 47.54; Observations 4220
- Maternal mortality: Mean 21.76; Median 6.50; Standard dev. 30.69; Observations 4220
- WBL index: Mean 58.75; Median 58.80; Standard dev. 19.28; Observations 7404
- Mobility: Mean 80.34; Median 100; Standard dev. 27.51; Observations 7404
- Workplace: Mean 42.49; Median 25; Standard dev. 34.40; Observations 7404
- Pay: Mean 46.72; Median 50; Standard dev. 30.45; Observations 7404
- Marriage: Mean 57.45; Median 60; Standard dev. 30.64; Observations 7404
- Parenthood: Mean 35.83; Median 40; Standard dev. 30.62; Observations 7404
- Entrepreneurship: Mean 71.65; Median 75; Standard dev. 22.33; Observations 7404
- Assets: Mean 73.08; Median 80; Standard dev. 28.57; Observations 7404
- Pension: Mean 62.27; Median 75; Standard dev. 27.50; Observations 7404
- Civil liberties: Mean 4.49; Median 5; Standard dev. 1.75; Observations 4188
- GDP: Mean 26.65; Median 24.48; Standard dev. 1.98; Observations 4157
- GDP per capita: Mean 8.35; Median 8.28; Standard dev. 1.45; Observations 4144
- Trade: Mean 78.50; Median 68.33; Standard dev. 46.50; Observations 3921
- FDI: Mean 3.91; Median 2.22; Standard dev. 10.37; Observations 3850
- Credit: Mean 48.07; Median 31.40; Standard dev. 44.68; Observations 2988
- Inflation: Mean 16.66; Median 4.60; Standard dev. 163.69; Observations 2980
- Agricultural employment: Mean 29.97; Median 23.10; Standard dev. 24.73; Observations 2880
- Population: Mean 16.26; Median 16.15; Standard dev. 1.46; Observations 4220

### Correlations with gender inequality index (GII) (Table A.3)
- Enrollment (female, primary): 0.31***
- Enrollment (female, secondary): 0.85***
- Enrollment (female, tertiary): 0.79***
- Women in parliament: -0.49***
- Labor force participation (female): -0.00
- Adolescent fertility: 0.82***
- Maternal mortality: 0.70***
- WBL index: -0.66***
- Civil liberties: -0.57***
- GDP per capita: -0.80***
- Trade: -0.33***
- FDI: -0.15***
- Credit: -0.71***
- Inflation: 0.05***
- Agricultural employment: 0.74***
- Notes: *** p<0.01, ** p<0.05, * p<0.1.

### Empirical specifications and figures described
- GDP per capita and conflicts (local projections in Section 3.3):
  - Specification reported:
    - log(GDP per capita_{c,t+p}) = β_p Political violence_{c,t-1} + θ_c + θ_t + ε_{c,t+p}
  - Figure 3 reports coefficient estimates of β_p for p = 0,1,...,9, as multiplied by 100 (to provide percentage points changes).
- Gender inequality around the onset of conflicts (Section 5.3):
  - Specification reported:
    - Gender inequality_{c,t} = Σ_{r=-4}^{4} β_r Political violence_{c,t-r} + θ_c + θ_t + ε_{c,t}
  - The paper documents coefficient estimates of β_r for r = −4, −3, ..., 0, ..., 3, 4 in Figure 4.

### Placebo tests
- Figure A.1: Placebo tests (t-statistics)
  - Notes: The results are based on equation 1. The chart reports the distribution of t-statistics from the 1000 placebo tests, where the placebo dummy variable is randomly assigned by keeping the overall probability of conflicts unchanged for each country. It also shows the t-statistic of the result in the first column of Table 1.

*Source: wpiea2025012-print-pdf - References*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025012-print-pdf.pdf_