## Appendix 1). Next are the combinations (inflation rate – exchange rate volatility) and (inflation

## Source details

**Canonical URL:** [Appendix 1). Next are the combinations (inflation rate – exchange rate volatility) and (inflation](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022096-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022096-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022096-print-pdf.pdf.json)

---

### Key empirical finding: optimal monetary policy outcomes and fragility
- The combination (inflation rate – unemployment rate) is identified as the “optimal” set of monetary policy outcomes that reduces the most the state of fragility.
- The combination (inflation rate – exchange rate volatility) is likely to be over-estimated because of high correlation between those variables; using highly correlated independent variables can bias estimated coefficients.

### Estimation strategy and model types
- Baseline approach: fixed effects and IV models; several additional covariates treated as potentially endogenous.
- Augmented specifications: opportunistic model (Equation 2), grievance model (Equation 3), and mix opportunity-grievance model (Equation 4).
- IV strategy: two-stage-least-square (2SLS) random-effects estimator.
- Instruments tested include trade openness (used to instrument natural resource rent), volume of exports, international reserves, lagged variables or differences (inflation, unemployment, real GDP per capita, real GDP growth and ODA).
- Instrument validation based on the Sargan test of over-identification.

### Main regression results — MEV as dependent variable
- Opportunity models (Table 5) — Inflation and unemployment coefficients (columns (1) to (5)):
  - Inflation: 0.35**, 0.26, 0.41**, 0.26, 0.91*** (standard errors: (0.17), (0.21), (0.19), (0.23), (0.22))
  - Unemployment: 0.05***, 0.03**, 0.04***, 0.04***, 0.07*** (standard errors: (0.01), (0.02), (0.01), (0.01), (0.01))
  - Observations: 2,064; 1,384; 2,044; 1,375; 1,963
  - Number of countries: 93; 89; 93; 89; 92
  - Sargan test p-value: 0.987; 0.864; 0.864; 0.696; 0.421
  - R2: 0.2805; 0.2213; 0.2246; 0.2473; 0.2770
- Grievance models (Table 6) — Inflation and unemployment coefficients (columns (1) to (5)):
  - Inflation: 0.44***, 0.40***, 0.35**, 0.51***, 0.38*** (standard errors: (0.12), (0.14), (0.14), (0.10), (0.14))
  - Unemployment: 0.02, 0.00, 0.01, 0.04***, 0.01 (standard errors: (0.01), (0.02), (0.02), (0.01), (0.02))
  - Observations: 1,305; 1,001; 1,001; 2,160; 947
  - Number of countries: 50; 47; 47; 99; 47
  - Sargan test p-value: 0.905; 0.710; 0.793; 0.382; 0.370
  - R2: 0.2822; 0.1944; 0.1979; 0.1470; 0.1501
- Merged grievance and opportunity models (Table 7) — Inflation and other coefficients (columns (1) to (5)):
  - Inflation: 0.39***, 0.58*, 0.37**, 0.48***, 0.46** (standard errors: (0.10), (0.33), (0.16), (0.17), (0.18))
  - Unemployment: 0.02 (standard error: (0.01)) reported in column (1)
  - Observations: 1,244; 1,690; 1,692; 651; 746
  - Number of countries: 50; 87; 87; 37; 38
  - Sargan test p-value: 0.187; 0.171; 0.127; 0.397; 0.203
  - R2 (between): 20.94; 22.42; 8.883; 25.33; 9.310
  - R2 (within): 2.187; 1.37e-05; 0.216; 1.731; 8.813
  - R2 (overall): 16.24; 3.624; 3.918; 8.630; 8.451

### Covariate patterns and signs
- Human capital (mean years of schooling, school enrolment) and ODA broadly negatively correlated with fragility.
- Natural resource rent, population size, and relative size of rural population positively associated with fragility.
- Real GDP per capita growth often displays the theoretically expected negative sign when included, sometimes rendering unemployment insignificant in merged specifications where both are present.

### Robustness analyses (methods and key outcomes)
- Alternative inflation measure:
  - Growth rate of the GDP deflator used instead of CPI; results remain similar— inflation (GDP deflator growth) mostly positively correlated with fragility.
- Sub-period analysis:
  - Sample split into 1980-1992 and 1993-2018 to check structural change around 1992. Results unchanged: inflation remains strongly correlated with fragility.
  - Note: unemployment data points available only for 1993-2018; findings on unemployment pertain to this period only.
- Dollarization split:
  - Share of firms’ dollar-denominated loans to total loans with threshold Q1 = 22.8 percent.
  - Inflation significant only in the group of countries that are not dollarized; unemployment significant in both sub-samples.
  - Two explanations offered: monetary policy credibility matters; existence of a dollarization threshold beyond which inflation is not a driver of fragility.
- Tobit censored model:
  - Accounts for dependent variable censoring; results align with continuous-model findings— inflation coefficient positive and significant; unemployment coefficients statistically significant with expected signs.
- Instrumenting inflation by monetary-policy-explained component:
  - Inflation component estimated via fixed-effects model where inflation is explained by growth of money supply; predicted inflation used as instrument in IV models (opportunity, grievance, unifying).
  - Results broadly consistent with hypothesis on impact of monetary policy outcomes on state fragility (Appendix 11).
- Fiscal policy inclusion:
  - Total government expenditure as percent of GDP added as exogenous or endogenous variable (first difference used as instrument when endogenous).
  - Coefficient on fiscal variable not significantly different from zero in opportunity models with more observations.
  - Findings on inflation and unemployment remain broadly unchanged (Appendix 12).
- Alternative fragility measure:
  - Uppsala Conflict Data Program fatalities from non-state violence used as dependent variable, with the inflation component instrumented by monetary-policy-explained inflation (Appendix 13).
  - Results: except for the grievance model, inflation matters in the opportunity model and to some extent in the unifying model.
  - In the unifying model, inflation becomes insignificant when polity2 (political regime proxy) is included; sample reduces to coverage of 37 countries where polity2 included.

### Methodological caveats and data limitations
- High correlation between some regressors (e.g., inflation and exchange rate volatility) can bias coefficient estimates.
- Data limitations prevented some analyses (e.g., checking for an inflation threshold effect per Hansen (1999) due to limited data).
- Potential persistent residual reverse causality effects may remain despite IV strategies; further work on instruments and measures of monetary policy effectiveness recommended.

### Policy implications and conclusions
- Monetary policy outcomes matter in fragile settings: maintaining relatively low inflation rates (price stability) and reducing unemployment (real stability) jointly yield the largest reduction in fragility.
- Single-objective monetary frameworks (e.g., strict inflation targeting) may not be appropriate in fragile contexts; central banks in fragile settings should balance nominal objectives with real objectives.
- Effectiveness of monetary policy transmission channels conditions the impact of monetary outcomes on fragility; improving transmission effectiveness is especially important in fragile environments.
- Further research recommended on:
  - Tensions between nominal and real objectives in monetary frameworks.
  - Identification of policy instruments and intermediate targets that enable central bankers in fragile settings to achieve price stability.
  - Coordination between fiscal and monetary policies in contexts of fragility.
  - Improved instruments and measures of monetary policy effectiveness and expanded data coverage.

### Appendix A — Correlation structure (selected)
- MEV correlations (correlation coefficient with p-value in brackets):
  - MEV with SFI: 0.39 (0.00)
  - MEV with INFLATION: 0.14 (0.00)
  - MEV with GDPCAP: -0.12 (0.00)
  - MEV with D.GDPCAP: -0.09 (0.00)
  - MEV with RENT: 0.03 (0.12)
  - MEV with polity2: -0.06 (0.00)
  - MEV with EFINDEX: 0.12 (0.00)
  - MEV with EXVOL: 0.07 (0.00)
  - MEV with POP: 0.22 (0.00)
  - MEV with SCHYEAR: -0.15 (0.00)
  - MEV with SCHENROLL: -0.12 (0.00)
  - MEV with RURPOP: 0.12 (0.00)
  - MEV with ODA: -0.07 (0.00)
  - MEV with UR: -0.05 (0.00)
  - MEV with Deflator: 0.04 (0.01)

### Appendix B — Descriptive statistics (key variables)
- MEV: Mean 1.05; Median 0.00; S.D. 2.13; Min 0.00; Max 14.00; # Observations 4,038
- SFI: Mean 11.10; Median 11.00; S.D. 5.13; Min 0.00; Max 25.00; # Observations 1,224
- Inflation: Mean 16.61; Median 7.10; S.D. 41.41; Min -129.94; Max 648.42; # Observations 3,743
- Real GDP per capita growth: Mean 1.52; Median 2.05; S.D. 6.85; Min -104.96; Max 87.70; # Observations 3,792
- Natural resources rent: Mean 9.76; Median 5.87; S.D. 11.22; Min 0.00; Max 86.45; # Observations 3,759
- polity2: Mean 0.86; Median 3.00; S.D. 6.55; Min -10.00; Max 10.00; # Observations 1,893
- EFINDEX: Mean 0.51; Median 0.56; S.D. 0.26; Min 0.01; Max 0.89; # Observations 3,271
- Exchange rate volatility: Mean 5.62; Median 2.70; S.D. 12.58; Min 0.00; Max 283.23; # Observations 3,889
- Population size (million): Mean 45.09; Median 9.94; S.D. 158.66; Min 0.23; Max 1392.73; # Observations 4,043
- Mean years of schooling: Mean 5.45; Median 5.20; S.D. 3.09; Min 0.00; Max 12.80; # Observations 3,030
- School enrollment: Mean 51.88; Median 49.19; S.D. 29.18; Min 2.48; Max 132.82; # Observations 2,683
- Rural population: Mean 55.88; Median 57.24; S.D. 20.99; Min 0.00; Max 95.66; # Observations 4,050
- ODA (% of GNI): Mean 6.80; Median 3.74; S.D. 9.23; Min -0.64; Max 94.95; # Observations 3,669
- Unemployment rate: Mean 7.53; Median 5.67; S.D. 6.04; Min 0.30; Max 37.98; # Observations 2,992
- GDP Deflator (D): Mean -0.04; Median -0.02; S.D. 1.06; Min -6.63; Max 9.68; # Observations 3,218

### Appendix D — Main estimation results: consistent patterns across model families (selected)
- Inflation:
  - Positive and statistically significant coefficients in multiple specifications across sub-periods and model types (examples: 0.89***; 0.72***; 0.41**; 0.78***; 0.54**; 0.63**; Tobit Opportunity models: 0.73***; 0.76***; 0.83***; 0.70***; Tobit Grievance models: 1.45***; 1.25***; 1.57***; 2.07***).
  - Inflation (t-1) also positive and significant in several Tobit specifications (examples: 0.80***; 0.75***; 0.60***; 0.70***; 0.49*; 1.15***).
- Unemployment:
  - Positive and often significant (examples: 0.07*** in one sub-period; Tobit Opportunity: 0.18***; 0.16***; Tobit Grievance: 0.14***; unemployment (t-1) 0.12***).
- Real GDP per capita growth:
  - Frequently negative and statistically significant in many specifications (examples: -3.40***; -3.30*** in sub-period regressions; Tobit Opportunity: -6.09***; -6.21***; -5.85***; Unified Tobit: -8.93***; -9.80***; -7.76***; -7.25***).
- Population size:
  - Generally positive and significant across models (examples: 0.58; 0.59; 0.40*; 0.25**; Tobit: 0.78***; 0.87***; 0.79***; 0.92***; unified Tobit: 0.72**; 0.77***; 1.62***; 1.60***; 1.54***).
- Rural population:
  - Positive and frequently statistically significant (examples: 0.05**; 0.05**; 0.04**; 0.02***; Tobit: 0.08***; 0.09***; 0.09***; unified Tobit: 0.11***; 0.10***; 0.05**).
- Mean years of schooling / School enrolment:
  - Higher schooling associated with lower MEV in many specifications (examples: mean years of schooling -0.29***; -0.25***; -0.28***; unified: -0.16**; school enrolment -1.54***; -2.19***; -1.67***; unified columns show -2.23***).
- Natural resource rent:
  - Mixed effects across specifications: positive and significant in some (examples: 1.71***; 1.96***; 1.28**) but negative and significant in other Tobit and unified specifications (examples: -0.62**; -0.81***; -0.63**; unified: -0.31; -0.75**; -0.71*).
- Ethnic fractionalization:
  - Positive and significant in several grievance/combined models (examples: 5.67**; 5.96***; 5.11**; unified: 4.40**; 4.20**), but not universally significant across all specifications.
- polity2 (political regime):
  - Negative and sometimes significant in some specifications (examples: -0.02*; -0.02* in sub-periods; unified Tobit shows polity2 -0.11** in one column; opportunity/grievance tables show -0.03** in multiple columns).

### Appendix E — Subsample and heterogeneity findings (selected)
- By sub-period (1980-1992 vs 1993-2018):
  - Inflation coefficients: 0.89*** (1980-1992), 0.72*** (1993-2018) and other period splits show continued significance in many columns.
  - Real GDP per capita growth: coefficients vary; some sub-periods show large negative and significant effects (examples: -3.40***; -3.30***).
- By level of dollarization (Above 22.8 percent vs Below or equal to 22.8 percent):
  - Inflation effects differ by dollarization: some columns show insignificant or negative coefficients (example: -0.02) while others show large positive and significant coefficients (examples: 3.42**; 8.53***; 9.73***).
  - Unemployment effect stronger in some high-dollarization subsamples (examples: 0.13***; 0.19***).
  - Population size remains positive and significant across dollarization splits (examples: 0.61***; 0.21**; 0.25**; 0.62***; 0.76***; 5.84*).
  - School enrolment has negative and sometimes significant coefficients in certain dollarization subsamples (examples: -0.65*; -2.04**).

### Appendix F — Tobit model diagnostics and variance components (selected)
- Tobit Opportunity models (multiple columns): sigma_u estimates range from 3.78*** to 4.28***; sigma_e estimates range from 2.14*** to 2.77***.
- Tobit Grievance models (multiple columns): sigma_u estimates range from 4.04*** to 5.86***; sigma_e estimates range from 2.45*** to 2.99***.
- Tobit Grievance-Opportunity Unifying models: sigma_u ranges (examples: 4.41***; 4.16***; 4.32***; 4.13***; 3.91***); sigma_e ranges (examples: 2.36***; 2.30***; 2.31***; 2.25***; 1.97***).

### Appendix G — Selected model fit and test outcomes (examples)
- Sargan test p-values reported across models (examples): 0.884; 0.839; 0.0612; 0.722; 0.299; 0.460.
- R2 examples: 0.1478; 0.1495; 0.09527; 0.1961.
- Log likelihoods reported for Tobit and unified models (examples): -2152; -2017; -1892; -1636; -950.5; -1255.

### Appendix 13: Opportunity, Grievance and Unifying Models – Dependent Variable: Number of Fatalities (selected notes)
- Dependent variable: Number of Fatalities (as indicated by appendix title).
- Exchange rate volatility definition:
  - "The exchange rate volatility is computed as the annual standard deviation of monthly nominal exchange rate divided by the monthly average of nominal exchange rate. It is basically a coefficient of variation."
  - "Data on exchange rates are extracted from the IFS, and it refers to the amount of local currency against US$ 1."

### Appendix 15 — Monetary policy framework, 2019 (selected entries)
- Exchange rate anchors and monetary frameworks (selected entries preserved exactly as in source):
  - "No separate legal tender: Timor-Leste Kosovo Montenegro"
  - "Currency board: Djibouti"
  - "Conventional peg: Eritrea Iraq Comoros Cameroon Central African Rep. Chad Gabon Congo, Rep Mali Burkina Faso Equatorial Guinea Côte d’Ivoire Guinea-Bissau Niger Togo Libya Nepal Solomon Islands"
  - "Stabilized arrangement: lLebanon North Macedonia Democratic Rep. of the Congo"
  - "Crawling peg: Honduras Nicaragua"
  - "Crawl-like arrangement: Liberia (7/18)"
  - "Other managed arrangement: Cambodia Syria Afghanistan Myanmar Sierra Leone The Gambia Venezuela"
  - "Floating: Belarus Madagascar Zimbabwe Albania Mozambique7 Zambia"
  - "Free floating: Somalia11"
- "Source: AREAER database."
- Selected footnote texts preserved exactly where shown (examples include notes 1, 5, 6, 7, 11).

### Appendix 16 — Descriptive statistics: World Bank classification of Fragile Countries (2005-2018) (selected)
- MEV: Mean 1.0, Median 0.0, S.D. 1.8, Min 0.0, Max 6.0, # Observations 495
- SFI: Mean 13.7, Median 14.0, S.D. 4.7, Min 3.0, Max 22.0, # Observations 198
- Inflation: Mean 0.1, Median 0.1, S.D. 0.1, Min -1.3, Max 0.9, # Observations 488
- GDPCAP: Mean 1974.0, Median 1199.8, S.D. 2112.9, Min 210.8, Max 9675.4, # Observations 488
- Exchange rate volatility: Mean 4.2, Median 2.2, S.D. 15.7, Min 0.0, Max 283.2, # Observations 477
- Unemployment rate: Mean 6.2, Median 4.6, S.D. 5.0, Min 0.4, Max 20.5, # Observations 484

### Appendix 17 — Descriptive statistics: World Bank classification of Non-Fragile Countries (2005-2018) (selected)
- MEV: Mean 0.5, Median 0.0, S.D. 1.4, Min 0.0, Max 7.0, # Observations 1,026
- SFI: Mean 8.8, Median 9.0, S.D. 4.2, Min 0.0, Max 21.0, # Observations 516
- Inflation: Mean 0.1, Median 0.1, S.D. 0.2, Min -0.3, Max 6.5, # Observations 998
- GDPCAP: Mean 4172.9, Median 3224.0, S.D. 3482.9, Min 233.9, Max 20533.0, # Observations 1,018
- Exchange rate volatility: Mean 3.2, Median 2.2, S.D. 4.4, Min 0.0, Max 63.2, # Observations 997
- Unemployment rate: Mean 7.7, Median 5.9, S.D. 5.8, Min 0.3, Max 30.8, # Observations 1,026

*Source: wpiea2022096-print-pdf - Appendix 1). Next are the combinations (inflation rate – exchange rate volatility) and (inflation*

### Appendix 1). Next are the combinations (inflation rate – exchange rate volatility) and (inflation

### wpiea2022096-print-pdf - Appendix 1). Next are the combinations (inflation rate – exchange rate volatility) and (inflation

### Key empirical finding: optimal monetary policy outcomes and fragility
- The combination (inflation rate – unemployment rate) is identified as the “optimal” set of monetary policy outcomes that reduces the most the state of fragility.
- The combination (inflation rate – exchange rate volatility) is likely to be over-estimated because of high correlation between those variables; using highly correlated independent variables can bias estimated coefficients.

### Estimation strategy and model types
- Baseline approach: fixed effects and IV models; several additional covariates treated as potentially endogenous.
- Augmented specifications: opportunistic model (Equation 2), grievance model (Equation 3), and mix opportunity-grievance model (Equation 4).
- IV strategy: two-stage-least-square (2SLS) random-effects estimator.
- Instruments tested include trade openness (used to instrument natural resource rent), volume of exports, international reserves, lagged variables or differences (inflation, unemployment, real GDP per capita, real GDP growth and ODA).
- Instrument validation based on the Sargan test of over-identification.

### Main regression results — MEV as dependent variable
- Opportunity models (Table 5) — Inflation and unemployment coefficients (columns (1) to (5)):
  - Inflation: 0.35**, 0.26, 0.41**, 0.26, 0.91*** (standard errors: (0.17), (0.21), (0.19), (0.23), (0.22))
  - Unemployment: 0.05***, 0.03**, 0.04***, 0.04***, 0.07*** (standard errors: (0.01), (0.02), (0.01), (0.01), (0.01))
  - Observations: 2,064; 1,384; 2,044; 1,375; 1,963
  - Number of countries: 93; 89; 93; 89; 92
  - Sargan test p-value: 0.987; 0.864; 0.864; 0.696; 0.421
  - R2: 0.2805; 0.2213; 0.2246; 0.2473; 0.2770
- Grievance models (Table 6) — Inflation and unemployment coefficients (columns (1) to (5)):
  - Inflation: 0.44***, 0.40***, 0.35**, 0.51***, 0.38*** (standard errors: (0.12), (0.14), (0.14), (0.10), (0.14))
  - Unemployment: 0.02, 0.00, 0.01, 0.04***, 0.01 (standard errors: (0.01), (0.02), (0.02), (0.01), (0.02))
  - Observations: 1,305; 1,001; 1,001; 2,160; 947
  - Number of countries: 50; 47; 47; 99; 47
  - Sargan test p-value: 0.905; 0.710; 0.793; 0.382; 0.370
  - R2: 0.2822; 0.1944; 0.1979; 0.1470; 0.1501
- Merged grievance and opportunity models (Table 7) — Inflation and other coefficients (columns (1) to (5)):
  - Inflation: 0.39***, 0.58*, 0.37**, 0.48***, 0.46** (standard errors: (0.10), (0.33), (0.16), (0.17), (0.18))
  - Unemployment: 0.02 (standard error: (0.01)) reported in column (1)
  - Observations: 1,244; 1,690; 1,692; 651; 746
  - Number of countries: 50; 87; 87; 37; 38
  - Sargan test p-value: 0.187; 0.171; 0.127; 0.397; 0.203
  - R2 (between): 20.94; 22.42; 8.883; 25.33; 9.310
  - R2 (within): 2.187; 1.37e-05; 0.216; 1.731; 8.813
  - R2 (overall): 16.24; 3.624; 3.918; 8.630; 8.451

### Covariate patterns and signs
- Human capital (mean years of schooling, school enrolment) and ODA broadly negatively correlated with fragility.
- Natural resource rent, population size, and relative size of rural population positively associated with fragility.
- Real GDP per capita growth often displays the theoretically expected negative sign when included, sometimes rendering unemployment insignificant in merged specifications where both are present.

### Robustness analyses (methods and key outcomes)
- Alternative inflation measure: growth rate of the GDP deflator used instead of CPI; results remain similar— inflation (GDP deflator growth) mostly positively correlated with fragility.
- Sub-period analysis: sample split into 1980-1992 and 1993-2018 to check structural change around 1992. Results unchanged: inflation remains strongly correlated with fragility.
  - Note: unemployment data points available only for 1993-2018; findings on unemployment pertain to this period only.
- Dollarization split: share of firms’ dollar-denominated loans to total loans with threshold Q1 = 22.8 percent.
  - Inflation significant only in the group of countries that are not dollarized; unemployment significant in both sub-samples.
  - Two explanations offered: monetary policy credibility matters; existence of a dollarization threshold beyond which inflation is not a driver of fragility.
- Tobit censored model: accounts for dependent variable censoring; results align with continuous-model findings— inflation coefficient positive and significant; unemployment coefficients statistically significant with expected signs.
- Instrumenting inflation by monetary-policy-explained component:
  - Inflation component estimated via fixed-effects model where inflation is explained by growth of money supply; predicted inflation used as instrument in IV models (opportunity, grievance, unifying).
  - Results broadly consistent with hypothesis on impact of monetary policy outcomes on state fragility (Appendix 11).
- Fiscal policy inclusion:
  - Total government expenditure as percent of GDP added as exogenous or endogenous variable (first difference used as instrument when endogenous).
  - Coefficient on fiscal variable not significantly different from zero in opportunity models with more observations.
  - Findings on inflation and unemployment remain broadly unchanged (Appendix 12).
- Alternative fragility measure: Uppsala Conflict Data Program fatalities from non-state violence used as dependent variable, with the inflation component instrumented by monetary-policy-explained inflation (Appendix 13).
  - Results: except for the grievance model, inflation matters in the opportunity model and to some extent in the unifying model.
  - In the unifying model, inflation becomes insignificant when polity2 (political regime proxy) is included; sample reduces to coverage of 37 countries where polity2 included.

### Methodological caveats and data limitations
- High correlation between some regressors (e.g., inflation and exchange rate volatility) can bias coefficient estimates.
- Data limitations prevented some analyses (e.g., checking for an inflation threshold effect per Hansen (1999) due to limited data).
- Potential persistent residual reverse causality effects may remain despite IV strategies; further work on instruments and measures of monetary policy effectiveness recommended.

### Policy implications and conclusions
- Monetary policy outcomes matter in fragile settings: maintaining relatively low inflation rates (price stability) and reducing unemployment (real stability) jointly yield the largest reduction in fragility.
- Single-objective monetary frameworks (e.g., strict inflation targeting) may not be appropriate in fragile contexts; central banks in fragile settings should balance nominal objectives with real objectives.
- Effectiveness of monetary policy transmission channels conditions the impact of monetary outcomes on fragility; improving transmission effectiveness is especially important in fragile environments.
- Further research recommended on:
  - Tensions between nominal and real objectives in monetary frameworks.
  - Identification of policy instruments and intermediate targets that enable central bankers in fragile settings to achieve price stability.
  - Coordination between fiscal and monetary policies in contexts of fragility.
  - Improved instruments and measures of monetary policy effectiveness and expanded data coverage.

*Source: wpiea2022096-print-pdf - Appendix 1). Next are the combinations (inflation rate – exchange rate volatility) and (inflation*

### 104. University of Massachusetts at Amherst: Political Economy Research Institute.

### 104. University of Massachusetts at Amherst: Political Economy Research Institute.

### Bibliographic and thematic context
- Selected references cited (authors and works as listed): Fearon & Laitin (2003); Feeny, Posso, & Regan-Beasley (2015); FitzGerald (1997); Garfinkel (1990); Gurr (1970); Gurr & Moore (1997); Gutiérrez et al. (2011); Homer-Dixon (1994); Hanke (2002); Hirshleifer (1988, 1989); Honohan & O’Connell (1997); Lopes, Hamdok, & Elhiraika (2017); Loungani & Swagel (2001); Mecagni et al. (2015); McGuirk & Burke (2017); Naudé, Santos-Paulino, & McGillivray (2011); Pinto (1989); Skaperdas (1992); Stewart (2003); World Bank (2011).

### Appendix A — Correlation structure (selected)
- MEV correlations (correlation coefficient with p-value in brackets):
  - MEV with SFI: 0.39 (0.00)
  - MEV with INFLATION: 0.14 (0.00)
  - MEV with GDPCAP: -0.12 (0.00)
  - MEV with D.GDPCAP: -0.09 (0.00)
  - MEV with RENT: 0.03 (0.12)
  - MEV with polity2: -0.06 (0.00)
  - MEV with EFINDEX: 0.12 (0.00)
  - MEV with EXVOL: 0.07 (0.00)
  - MEV with POP: 0.22 (0.00)
  - MEV with SCHYEAR: -0.15 (0.00)
  - MEV with SCHENROLL: -0.12 (0.00)
  - MEV with RURPOP: 0.12 (0.00)
  - MEV with ODA: -0.07 (0.00)
  - MEV with UR: -0.05 (0.00)
  - MEV with Deflator: 0.04 (0.01)
- Note: P-values of significance tests of correlation coefficients are in brackets (below).

### Appendix B — Descriptive statistics (key variables)
- MEV: Mean 1.05; Median 0.00; S.D. 2.13; Min 0.00; Max 14.00; # Observations 4,038
- SFI: Mean 11.10; Median 11.00; S.D. 5.13; Min 0.00; Max 25.00; # Observations 1,224
- Inflation: Mean 16.61; Median 7.10; S.D. 41.41; Min -129.94; Max 648.42; # Observations 3,743
- Real GDP per capita growth: Mean 1.52; Median 2.05; S.D. 6.85; Min -104.96; Max 87.70; # Observations 3,792
- Natural resources rent: Mean 9.76; Median 5.87; S.D. 11.22; Min 0.00; Max 86.45; # Observations 3,759
- polity2: Mean 0.86; Median 3.00; S.D. 6.55; Min -10.00; Max 10.00; # Observations 1,893
- EFINDEX: Mean 0.51; Median 0.56; S.D. 0.26; Min 0.01; Max 0.89; # Observations 3,271
- Exchange rate volatility: Mean 5.62; Median 2.70; S.D. 12.58; Min 0.00; Max 283.23; # Observations 3,889
- Population size (million): Mean 45.09; Median 9.94; S.D. 158.66; Min 0.23; Max 1392.73; # Observations 4,043
- Mean years of schooling: Mean 5.45; Median 5.20; S.D. 3.09; Min 0.00; Max 12.80; # Observations 3,030
- School enrollment: Mean 51.88; Median 49.19; S.D. 29.18; Min 2.48; Max 132.82; # Observations 2,683
- Rural population: Mean 55.88; Median 57.24; S.D. 20.99; Min 0.00; Max 95.66; # Observations 4,050
- ODA (% of GNI): Mean 6.80; Median 3.74; S.D. 9.23; Min -0.64; Max 94.95; # Observations 3,669
- Unemployment rate: Mean 7.53; Median 5.67; S.D. 6.04; Min 0.30; Max 37.98; # Observations 2,992
- GDP Deflator (D): Mean -0.04; Median -0.02; S.D. 1.06; Min -6.63; Max 9.68; # Observations 3,218

### Appendix C — Unit-root testing
- Appendix 3: Fisher Panel Unit Root Tests (tables presented; no numeric results excerpted here).
- Appendix 4: Im-Pesaran-Shin Unit Root Tests (tables presented; no numeric results excerpted here).

### Appendix D — Main estimation results: consistent patterns across model families
- Inflation:
  - Positive and statistically significant coefficients in multiple specifications across sub-periods and model types (e.g., 0.89***; 0.72***; 0.41**; 0.78***; 0.54**; 0.63**; Tobit Opportunity models: 0.73***; 0.76***; 0.83***; 0.70***; Tobit Grievance models: 1.45***; 1.25***; 1.57***; 2.07***).
  - Inflation (t-1) also positive and significant in several Tobit specifications (e.g., 0.80***; 0.75***; 0.60***; 0.70***; 0.49*; 1.15***).
- Unemployment:
  - Positive and often significant (e.g., 0.07*** in one sub-period; Tobit Opportunity: 0.18***; 0.16***; Tobit Grievance: 0.14***; unemployment (t-1) 0.12***).
- Real GDP per capita growth:
  - Frequently negative and statistically significant in many specifications (e.g., -3.40***; -3.30*** in sub-period regressions; Tobit Opportunity: -6.09***; -6.21***; -5.85***; Unified Tobit: -8.93***; -9.80***; -7.76***; -7.25***).
- Population size:
  - Generally positive and significant across models (examples: 0.58; 0.59; 0.40*; 0.25**; Tobit: 0.78***; 0.87***; 0.79***; 0.92***; unified Tobit: 0.72**; 0.77***; 1.62***; 1.60***; 1.54***).
- Rural population:
  - Positive and frequently statistically significant (examples: 0.05**; 0.05**; 0.04**; 0.02***; Tobit: 0.08***; 0.09***; 0.09***; unified Tobit: 0.11***; 0.10***; 0.05**).
- Mean years of schooling / School enrolment:
  - Higher schooling associated with lower MEV in many specifications: Mean years of schooling negative and significant in Tobit Opportunity/Unified models (e.g., -0.29***; -0.25***; -0.28***; unified: -0.16**). School enrolment negative and significant in several Tobit and unified models (e.g., -1.54***; -2.19***; -1.67***; unified columns show -2.23***).
- Natural resource rent:
  - Mixed effects across specifications: positive and significant in some (e.g., 1.71***; 1.96***; 1.28**) but negative and significant in other Tobit and unified specifications (e.g., -0.62**; -0.81***; -0.63**; unified: -0.31; -0.75**; -0.71*).
- Ethnic fractionalization:
  - Positive and significant in several grievance/combined models (examples: 5.67**; 5.96***; 5.11**; unified: 4.40**; 4.20**), but not universally significant across all specifications.
- polity2 (political regime):
  - Negative and sometimes significant in some specifications (examples: -0.02*; -0.02* in sub-periods; unified Tobit shows polity2 -0.11** in one column; opportunity/grievance tables show -0.03** in multiple columns).

### Appendix E — Subsample and heterogeneity findings (selected)
- By sub-period (1980-1992 vs 1993-2018):
  - Inflation coefficients: 0.89*** (1980-1992), 0.72*** (1993-2018) and other period splits show continued significance in many columns.
  - Real GDP per capita growth: coefficients vary; some sub-periods show large negative and significant effects (e.g., -3.40***; -3.30***).
- By level of dollarization (Above 22.8 percent vs Below or equal to 22.8 percent):
  - Inflation effects differ by dollarization: some columns show insignificant or negative coefficients (e.g., -0.02) while others show large positive and significant coefficients (e.g., 3.42**; 8.53***; 9.73***).
  - Unemployment effect stronger in some high-dollarization subsamples (e.g., 0.13***; 0.19***).
  - Population size remains positive and significant across dollarization splits (e.g., 0.61***; 0.21**; 0.25**; 0.62***; 0.76***; 5.84*).
  - School enrolment has negative and sometimes significant coefficients in certain dollarization subsamples (e.g., -0.65*; -2.04**).

### Appendix F — Tobit model diagnostics and variance components (selected)
- Tobit Opportunity models (multiple columns): sigma_u estimates range from 3.78*** to 4.28***; sigma_e estimates range from 2.14*** to 2.77***.
- Tobit Grievance models (multiple columns): sigma_u estimates range from 4.04*** to 5.86***; sigma_e estimates range from 2.45*** to 2.99***.
- Tobit Grievance-Opportunity Unifying models: sigma_u ranges (e.g., 4.41***; 4.16***; 4.32***; 4.13***; 3.91***); sigma_e ranges (e.g., 2.36***; 2.30***; 2.31***; 2.25***; 1.97***).

### Appendix G — Selected model fit and test outcomes (examples)
- Sargan test p-values reported across models (examples): 0.884; 0.839; 0.0612; 0.722; 0.299; 0.460.
- R2 examples: 0.1478; 0.1495; 0.09527; 0.1961.
- Log likelihoods reported for Tobit and unified models (examples): -2152; -2017; -1892; -1636; -950.5; -1255.

*Source: wpiea2022096-print-pdf - 104. University of Massachusetts at Amherst: Political Economy Research Institute.*

### Appendix 13: Opportunity, Grievance and Unifying Models – Dependent Variable: Number

### Appendix 13: Opportunity, Grievance and Unifying Models – Dependent Variable: Number of Fatalities

### Model focus and data notes
- Dependent variable: Number of Fatalities (as indicated by appendix title).
- Exchange rate volatility definition:
  - "The exchange rate volatility is computed as the annual standard deviation of monthly nominal exchange rate divided by the monthly average of nominal exchange rate. It is basically a coefficient of variation."
  - "Data on exchange rates are extracted from the IFS, and it refers to the amount of local currency against US$ 1."

### Appendix 15 — Monetary policy framework, 2019 (high-level classification highlights)
- Exchange rate anchors and monetary frameworks (selected entries and annotations preserved exactly as in source):
  - "No separate legal tender: Timor-Leste Kosovo Montenegro"
  - "Currency board: Djibouti"
  - "Conventional peg: Eritrea Iraq Comoros Cameroon Central African Rep. Chad Gabon Congo, Rep Mali Burkina Faso Equatorial Guinea Côte d’Ivoire Guinea-Bissau Niger Togo Libya Nepal Solomon Islands"
  - "Stabilized arrangement: lLebanon North Macedonia Democratic Rep. of the Congo"
  - "Crawling peg: Honduras Nicaragua"
  - "Crawl-like arrangement: Liberia (7/18)"
  - "Pegged exchange rate within horizontal bands: (no specific country list in excerpt)"
  - "Other managed arrangement: Cambodia Syria Afghanistan Myanmar Sierra Leone The Gambia Venezuela"
  - "Floating: Belarus Madagascar Zimbabwe Albania Mozambique7 Zambia"
  - "Free floating: Somalia11"
- Annotations and footnotes preserved exactly:
  - "Source: AREAER database."
  - Notes including numeric footnotes 1 through 11 and their texts as presented (examples):
    - "1 Includes countries that have no explicitly stated nominal anchor, but rather monitor various indicators in conducting monetary policy."
    - "5 The country maintains a de facto exchange rate anchor to the US dollar."
    - "6 The country maintains a de facto exchange rate anchor to the euro."
    - "7 The central bank is in transition toward inflation targeting."
    - "11 Currently the Central Bank of Somalia does not have a monetary policy framework."

### Appendix 16 — Descriptive statistics of key variables: World Bank classification of Fragile Countries (2005-2018)
- Observations and summary statistics (variables listed with Mean, Median, S.D., Min, Max, # Observations) — entries preserved exactly as in source:
  - MEV: Mean 1.0, Median 0.0, S.D. 1.8, Min 0.0, Max 6.0, # Observations 495
  - SFI: Mean 13.7, Median 14.0, S.D. 4.7, Min 3.0, Max 22.0, # Observations 198
  - Inflation: Mean 0.1, Median 0.1, S.D. 0.1, Min -1.3, Max 0.9, # Observations 488
  - D.GDP: Mean 0.0, Median 0.0, S.D. 0.1, Min -0.5, Max 0.2, # Observations 487
  - GDPCAP: Mean 1974.0, Median 1199.8, S.D. 2112.9, Min 210.8, Max 9675.4, # Observations 488
  - Real GDP per capita growth: Mean 0.0, Median 0.0, S.D. 0.1, Min -0.5, Max 0.2, # Observations 487
  - Natural resources rent: Mean 15.4, Median 12.1, S.D. 14.2, Min 0.0, Max 63.9, # Observations 446
  - D.RENT: Mean 0.0, Median 0.0, S.D. 0.3, Min -2.3, Max 2.3, # Observations 445
  - polity2: Mean 0.2, Median 2.0, S.D. 6.0, Min -9.0, Max 9.0, # Observations 198
  - EFINDEX: Mean 0.6, Median 0.7, S.D. 0.3, Min 0.1, Max 0.9, # Observations 292
  - Exchange rate volatility: Mean 4.2, Median 2.2, S.D. 15.7, Min 0.0, Max 283.2, # Observations 477
  - Population size (million): Mean 18.8, Median 9.5, S.D. 29.6, Min 0.5, Max 195.9, # Observations 488
  - Mean years of schooling: Mean 5.1, Median 4.6, S.D. 2.3, Min 1.5, Max 12.5, # Observations 484
  - School enrollment: Mean 49.6, Median 45.6, S.D. 21.1, Min 13.0, Max 111.9, # Observations 237
  - Rural population: Mean 54.1, Median 59.2, S.D. 22.7, Min 0.0, Max 90.6, # Observations 495
  - ODA: Mean 9.1, Median 5.5, S.D. 11.6, Min -0.5, Max 92.1, # Observations 477
  - Unemployment rate: Mean 6.2, Median 4.6, S.D. 5.0, Min 0.4, Max 20.5, # Observations 484
  - GDP Deflator (D): Mean 0.0, Median 0.0, S.D. 1.2, Min -5.8, Max 5.2, # Observations 381

### Appendix 17 — Descriptive statistics of key variables: World Bank classification of Non-Fragile Countries (2005-2018)
- Observations and summary statistics (variables listed with Mean, Median, S.D., Min, Max, # Observations) — entries preserved exactly as in source:
  - MEV: Mean 0.5, Median 0.0, S.D. 1.4, Min 0.0, Max 7.0, # Observations 1026
  - SFI: Mean 8.8, Median 9.0, S.D. 4.2, Min 0.0, Max 21.0, # Observations 516
  - Inflation: Mean 0.1, Median 0.1, S.D. 0.2, Min -0.3, Max 6.5, # Observations 998
  - D.GDP: Mean 0.0, Median 0.0, S.D. 0.1, Min -1.0, Max 0.8, # Observations 1016
  - GDPCAP: Mean 4172.9, Median 3224.0, S.D. 3482.9, Min 233.9, Max 20533.0, # Observations 1018
  - Real GDP per capita growth: Mean 0.0, Median 0.0, S.D. 0.1, Min -1.0, Max 0.8, # Observations 1016
  - Natural resources rent: Mean 9.7, Median 5.2, S.D. 11.6, Min 0.1, Max 68.8, # Observations 945
  - D.RENT: Mean 0.0, Median 0.0, S.D. 0.3, Min -1.8, Max 4.0, # Observations 941
  - polity2: Mean 3.3, Median 6.0, S.D. 6.0, Min -9.0, Max 10.0, # Observations 516
  - EFINDEX: Mean 0.5, Median 0.5, S.D. 0.2, Min 0.0, Max 0.9, # Observations 612
  - Exchange rate volatility: Mean 3.2, Median 2.2, S.D. 4.4, Min 0.0, Max 63.2, # Observations 997
  - Population (million): Mean 67.9, Median 14.5, S.D. 212.4, Min 0.5, Max 1392.7, # Observations 1026
  - Mean years of schooling: Mean 7.2, Median 7.4, S.D. 2.8, Min 0.0, Max 12.8, # Observations 998
  - School enrollment: Mean 72.4, Median 81.0, S.D. 26.2, Min 9.6, Max 132.8, # Observations 743
  - Rural population: Mean 48.2, Median 46.2, S.D. 19.8, Min 8.1, Max 84.9, # Observations 1026
  - ODA: Mean 3.2, Median 1.5, S.D. 4.2, Min -0.3, Max 27.5, # Observations 986
  - Unemployment rate: Mean 7.7, Median 5.9, S.D. 5.8, Min 0.3, Max 30.8, # Observations 1026
  - GDP Deflator (D): Mean 0.0, Median 0.0, S.D. 0.9, Min -4.3, Max 3.7, # Observations 920

*Do Monetary Policy Outcomes Promote Stability in Fragile Settings? Working Paper No. WP/2022/095 — Appendix 13 (extracted pages).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022096-print-pdf.pdf_