## 6. Nexus of Growth and Government Effectiveness

## Source details

**Canonical URL:** [6. Nexus of Growth and Government Effectiveness](https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021133-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2021/english/wpiea2021133-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2021/english/wpiea2021133-print-pdf.pdf.json)

---

### Introduction and empirical approach
- Fragile and conflict-affected States (FCS) defined as countries trapped in cycles of low administrative capacity, political instability, conflict, and weak economic performance.
- Study objectives:
  - (i) factors that push a country into fragility;
  - (ii) factors that help FCS exit fragility;
  - (iii) factors that help former FCS stay out of fragility and sustain good economic performance.
- Empirical methods:
  - Event studies;
  - Synthetic control method (SCM) focusing on real GDP per capita;
  - Logit model estimating conditional probability of entry/exit.
- Sample and classification:
  - Countries spanning 1977–2018; final samples include 128 countries.
  - Fragility if: rolling average CPIA (last three years) < 3.2; OR UN/regional PKO/PBO in most recent three years; OR conflict = top 25 percentile of deaths as percent of population.
  - Identified events (since 1981): 83 entry events; 52 exit events.
  - Of entry events: about 1/3 exited from fragility within 10 years; 2/3 stayed fragile after 10 years.
  - Of exit events: 30 percent returned to fragility within 10 years.
  - Counterfactuals (after 1995): 45 entry counterfactuals; 42 exit counterfactuals.
  - “Pivotal moments” compiled as crisis episodes or change in executive power, defined between T-2 and T+2.

### Event study results (T-4 to T-2 comparisons)
- Political institutions and competition:
  - Political competition (Polity IV) affects both entry into fragility and sustained exit from fragility (medians statistically different by Mood’s median test).
  - Executive constraints do not differentiate identified events from counterfactuals before events; sustained exit cases have stronger executive constraints than those returning to fragility after exit (median test marginally significant at 10 percent for T+2–T+4).
- Social spending and human development:
  - Health and education spending and Human Development Index (HDI) help sustained exit:
    - Health and education spending significant at 1 percent by median test.
    - Exit events have statistically stronger HDI than counterfactuals (median test).
  - Robustness: measuring social spending as percent of overall spending suggests preserving health spending could help early exit (sample too small for strong statistical tests).
- Pivotal moments and conflict:
  - Share of exit events with pivotal moments around exit: about 20 percent vs. 10 percent in counterfactuals (statistically significant at 1 percent).
  - For sustained exit and early exit: shares about 25 percent vs. 5 percent and about 25 percent vs. 10 percent (statistically significant at 1 percent).
  - Share of exit events with conflict: 30 percent vs. 40 percent in counterfactuals (statistically significant at 5 percent).
- Robustness checks:
  - Dropping conflict episodes: political competition and HDI results broadly hold; median differences for health and education spending become smaller; inference limited by small sample.

### Synthetic Control Method (real GDP per capita) — economic costs and benefits
- Cost of staying in fragility (median gaps vs synthetic control):
  - At T+3: median gap ≈ -12 percentage points.
  - At T+7: median gap ≈ -24 percentage points.
  - Quartile bands: T+3 = [-4, -23] percentage points; T+7 = [-11, -36] percentage points.
- Benefit of early exit (median deviations):
  - At T+3: countries that remained in fragility are about 8 percentage points lower than early-exit cases.
  - At T+7: difference grows to about 12 percentage points.
  - Statistical significance: differences significant at 10 percent level at T+3 and 5 percent at T+7 by Mood median test.
- Pivotal moments and sustained exit:
  - Sustained exit cases with pivotal moments outperform those without:
    - Average median deviation ≈ 9 percentage points during first 4 years and ≈ 11 percentage points during first 7 years relative to sustained exit without pivotal moments (differences significant at 5 percent).
  - Sustained exit with pivotal moments: systematically higher CPIA scores after exit and lower CPIA scores before exit (suggesting reform adoption during pivotal moments).

### Logit model (conditional probabilities and key coefficients)
- Sample sizes:
  - Entry logit: 85 observations (identified entry events + counterfactuals).
  - Exit logit: 69 observations (identified exit events + counterfactuals).
- Most robust determinants: real GDP per capita growth and government effectiveness.
- Implied probability effects (evaluated at means of other regressors):
  - Higher real GDP per capita growth reduces probability of entering fragility and raises probability of exiting fragility.
  - Stronger government effectiveness reduces probability of entry and raises probability of exit.
- Asymmetries and non-linearities:
  - Growth matters more for entry than for exit:
    - Example implied effect: probability of entry declines by 75 percentage points when growth rises from -5 percent to 5 percent; same change raises probability of exit by only 20 percentage points.
  - Government effectiveness effects are more symmetric and display a “hot” zone where marginal benefits from improvements are larger (non-linear effect concentrated around [-1.5, 1]).
  - Interaction interpretation:
    - Preventing a growth collapse is especially critical to avoid entering fragility for countries with intermediate government effectiveness (hot zone).
    - Improving government effectiveness matters more for sustaining exits and making exits robust to growth shocks; benefits of maintaining growth to avoid fragility diminish when government effectiveness is very weak (less than -1).

### Selected logit coefficients (high-level highlights from regression tables)
- Entry models (selected coefficients, standard errors in parentheses):
  - GDP per capita growth, year T-3 to T-1, average: coefficients range (examples) -47.2975*** (14.2282); -42.6886*** (13.3086); -43.8960*** (13.3384); -39.4549** (15.9038).
  - Government effectiveness, year T-1: several models show negative coefficients around -1.9832* (1.1207); -1.6427* (0.9413); -1.7839* (1.0237); -1.9712* (1.0664).
- Exit models (selected coefficients, standard errors in parentheses):
  - GDP per capita growth, year T-3 to T-1, average: positive coefficients around 13.1823* (7.1244); 13.5508* (7.1947); 15.9313* (8.9134); 19.6929** (8.7387).
  - Government effectiveness, year T-1: positive and significant coefficients such as 2.6118** (1.0294); 2.6675*** (0.9313); 3.8820*** (1.4297); 3.3446*** (1.1734).
- Significance notation used: *** p<0.01, ** p<0.05, * p<0.1.

### Policy implications and recommendations
- Core messages:
  - Persistence of fragility implies high hurdles to exit and high benefits to avoiding entry.
  - Nexus between growth/resilience and government effectiveness: avoiding growth collapses is crucial to prevent entries; improving government effectiveness is central to enable sustainable exits.
- Policy recommendations:
  1. Near-fragile countries should adopt timely and appropriate counter-cyclical policies to preserve economic activity.
     - Enhanced fiscal stimulus is needed to prevent a sharp output contraction, but its effect is weakened in countries with very weak government effectiveness.
     - International community should provide financing subject to appropriate governance safeguards and policy design.
     - In good times, build and maintain external and fiscal buffers through prudent policies to enable counter-cyclical responses when needed.
  2. Improve institutions and enhance political and social inclusion to sustain exits.
     - Protect social spending (health and education) to support political and social inclusion, which with good policies fosters growth and resilience.
     - Fiscal capacity building (revenue mobilization and PFM) complements institutional development and state building.
     - Economic gains from exiting fragility (per SCM) can finance further institution, human capital and infrastructure building.
  3. Seize “pivotal moments” to embrace reforms.
     - Pivotal moments create opportunities to reset citizen expectations and build trust; associated with stronger economic outcomes and policy/institutional improvements in sustained exit cases.
     - IFIs can provide technical assistance and capacity development during pivotal moments where political commitment to reform exists.

### Case studies — illustrative country findings
- Rwanda — findings and policy-relevant factors:
  - Political commitment, accountability, reform ownership, and international partner support helped sustain reforms after regaining political stability.
  - Pre-exit CPIA: about 2.2 in mid-1990s; post-exit CPIA: consistently above 3.5.
  - Tax-to-GDP ratio: increased from 9.7 percent in 2000 to over 13 percent on average during 2010-13.
  - Priority spending: increased from 4 percent of GDP in 1999 to 12-14 percent during 2008-12.
  - Poverty rate: declined from 75 percent (on average) in 1992-2000 to around 59 percent in 2010-16.
  - Role of external support: Fund-supported programs, HIPC and MDRI debt relief, IFI and donor technical support on revenue mobilization, PFM, monetary and exchange rate management.
  - Key reforms: tax administration modernization, tax policy (indirect tax rates), reinstituting budget process and PFM systems.
- Uganda — findings and policy-relevant factors:
  - Exit around mid-2000s as conflicts subsided; peace talks started in 2006.
  - CPIA: slightly less than 4 (well above threshold 3.2).
  - Tax revenue: increased from slightly below 9½ percent of GDP on average before peace talks to over 11 percent on average during the second half of the 2010s.
  - Poverty: share below poverty line declined from 65 percent in 2002 to 36 percent in 2012.
  - Key reforms: revenue-side reforms (eliminate exemptions, more progressive income tax, excise tax reform), tax administration improvements (high net worth unit, medium-sized taxpayer office, electronic tax services), PFM reforms (Treasury Single Account).
  - External support: IMF policy advice and three Policy Support Instrument programs (PSI); technical support from Fund and donors.
  - Note: some progress reversed after PSI programs ended in 2017 (e.g., tax exemptions increased; control of corruption index worsened; on-budget donor support declined).

*Source: IMF working paper chapter “6. Nexus of Growth and Government Effectiveness.”*

### REFERENCES _____________________________________________________________________________________ 30

### REFERENCES

### Figures
- 1. Event Studies: Political Competition and Executive Constraints  _______________________________ 16
- 2. Event Studies: Health and Education Spending and HDI  ______________________________________ 17
- 3. Event Studies: Health and Education Spending in percent of Overall Spending _______________ 18
- 4. GDP Per Capita Developments for Identified Events ___________________________________________ 21
- 5. The Determinants of Entry/Exit from Fragility __________________________________________________ 24

*Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021133-print-pdf.pdf*

### 6. Nexus of Growth and Government Effectiveness ______________________________________________ 26

### 6. Nexus of Growth and Government Effectiveness

### Introduction
- Fragile and conflict-affected States (FCS) are defined as countries trapped in cycles of low administrative capacity, political instability, conflict, and weak economic performance.
- Fragility is highly persistent with high risk of relapses; dynamics do not always move linearly (e.g., conflict → post-conflict → relapse).
- The “fragility trap” links slow-moving factors (institutional and policy weakness) and fast-moving factors (political conflict, exogenous shocks, economic/social instability).
- This study explores:
  - (i) factors that push a country into fragility;
  - (ii) factors that help FCS exit fragility; and
  - (iii) factors that help former FCS stay out of fragility and sustain good economic performance.
- Empirical approach:
  - Identify entry/exit events using a dataset (countries spanning 1979–2018 in one description; sample expanded to include countries with CPIA data back to late 1970s).
  - Use counterfactuals due to small numbers of events.
  - Apply three methodologies: event studies; synthetic control method (SCM); and a logit model.

### Literature review and conceptual framing
- Common characteristics of fragility: (i) institutional and policy implementation weakness; (ii) fractious political context; (iii) severe domestic resource constraints; (iv) high vulnerability to shocks.
- Operational definition used: WB/IMF approach based on CPIA ratings and presence of UN/regional peacekeeping/peace-building operations (PKO/PBO); extended in this paper by adding a conflict criterion based on deaths as percent of population.
- Determinants emphasized in literature:
  - Political instability, violence, insecurity, weak institutions, corruption.
  - Macroeconomic variables: income levels, economic growth; openness to trade.
  - Predetermined factors: country size, ethnic risk/diversity.
  - Social indicators: HDI, infant mortality, education.
- Policy implications from literature:
  - Strengthening institutions and governance is critical.
  - Fiscal institutions and fiscal space are associated with resilience.
  - Capacity building needs realistic tools, on-the-ground experts, adequate financing, and attention to political economy.

### Empirical strategy
- Challenges: high persistence of fragility yields few turning points; need large number of observations for standard econometrics.
- Construct counterfactuals: country-periods that should have entered/exited fragility (based on governance indicators) but did not.
- Three methods used for robustness:
  - Event studies across a wide range of indicators.
  - Synthetic control method (SCM) focusing on real GDP per capita.
  - Logit model estimating conditional probability of entry/exit.

### Data: turning points, pivotal moments, and samples
- Fragility classification in this paper: a country is fragile if:
  - its rolling average CPIA over the last three years is below 3.2; or
  - it has had a UN or regional PKO/PBO in the most recent three years; or
  - it has conflict defined as being in the top 25 percentile of all countries in terms of deaths as a proportion of the previous year’s total population.
- Sample expanded to include countries that used to be low income/lower middle income with CPIA data (1977–2018); final samples include 128 countries spanning from 1977 to 2018.
- Identified events (since 1981):
  - 83 entry events.
  - 52 exit events.
  - Of entry events: about 1/3 exited from fragility within 10 years; remaining 2/3 stayed fragile even after 10 years.
  - Of exit events: 30 percent came back to fragility within 10 years after exit.
- Counterfactuals (after 1995):
  - 45 entry counterfactuals.
  - 42 exit counterfactuals.
  - Entry counterfactual criterion: at least one of three WGI components (government effectiveness, regulatory quality, political stability) below 40th percentile but country not fragile in T to T+3.
  - Exit counterfactual criterion: at least one of the three WGI components greater than 40th percentile but country stays fragile in T to T+3.
- “Pivotal moments” compiled: crisis episodes (Laeven and Valencia, 2018) or change in executive power (Wilson, 2019); defined as moments between T-2 and T+2.

### Empirical results

#### A. Event study analysis (indicators compared T-4 to T-2)
Key findings:
- Political competition (Polity IV) affects both entry into fragility and sustained exit from fragility.
  - Identified events and counterfactuals have statistically-significant different medians by Mood’s median test.
  - Executive constraints do not differentiate identified events from counterfactuals before events, but sustained exit cases have stronger executive constraints than those returning to fragility after exit (median test marginally significant at 10 percent for T+2–T+4).
- Social spending (health and education) and Human Development Index (HDI) help sustained exit:
  - Health and education spending significant at 1 percent by median test.
  - Exit events have statistically stronger HDI than counterfactuals (median test).
- Pivotal moments are more associated with exit events:
  - Share of exit events with pivotal moments around the exit is about 20 percent vs. 10 percent in counterfactuals (statistically significant at 1 percent).
  - For sustained exit and early exit, shares about 25 percent vs. 5 percent and about 25 percent vs. 10 percent (statistically significant at 1 percent).
- Conflict has a lasting effect on exit likelihood:
  - Share of exit events with conflict is 30 percent vs. 40 percent in counterfactuals (statistically significant at 5 percent).
- Robustness checks:
  - Measuring social spending as percent of overall spending suggests preserving health spending could help early exit (sample too small for strong statistical tests).
  - Dropping conflict episodes: political competition and HDI results generally hold; differences in medians for health and education spending become smaller, but small sample limits inference.

#### B. Synthetic Control Method (SCM) — real GDP per capita
- SCM compares actual real GDP per capita to a synthetic control (counterfactual) at T+3 and T+7.
- Cost of staying in fragility:
  - Median gap of per capita GDP between countries staying in fragility and control group is about -12 percentage points after 3 years and -24 percentage points after 7 years.
  - Quartile bands: [-4, -23] percentage points at T+3 and [-11, -36] percentage points at T+7.
- Benefit of early exit:
  - Median deviation: countries that remained in fragility are about 8 percentage points lower at T+3 compared to early-exit cases; difference grows to about 12 percentage points at T+7.
  - Differences statistically significant: 10 percent level at T+3 and 5 percent at T+7 by Mood median test.
- Pivotal moments and sustained exit:
  - Sustained exit cases with pivotal moments show stronger real GDP per capita relative to control than those without pivotal moments.
  - Average median deviation for sustained exit with pivotal moments: about 9 percentage points during first 4 years and 11 percentage points during first 7 years relative to sustained exit without pivotal moments (differences significant at 5 percent).
  - Sustained exit cases with pivotal moments have systematically higher CPIA scores after exit and lower CPIA scores before exit (suggesting reform adoption during pivotal moments).

#### C. Logit model (conditional probability of entry/exit)
- Sample sizes:
  - Entry logit: 85 observations (identified entry events + counterfactuals).
  - Exit logit: 69 observations (identified exit events + counterfactuals).
- Given sample size constraints, regressions limited to 3 or 4 explanatory variables.
- Most robust determinants: real GDP per capita growth and government effectiveness.
- Implied probability effects (evaluated at means of other regressors):
  - Higher real GDP per capita growth reduces probability of entering fragility and raises probability of exiting fragility.
  - Stronger government effectiveness reduces probability of entry and raises probability of exit.
- Asymmetries:
  - Growth matters more for entry than for exit:
    - Example: implied probability of entry declines by 75 percentage points when growth rises from -5 percent to 5 percent; same change raises probability of exit by only 20 percentage points.
  - Government effectiveness effects are more symmetric and show a “hot” zone where improvements yield larger marginal benefits (non-linear effect concentrated in a domain around [-1.5, 1]).
  - Interaction interpretation:
    - Preventing a growth collapse is especially critical to avoid entering fragility for countries with some intermediate level of government effectiveness (hot zone).
    - Improving government effectiveness matters more for sustaining exits and making exits robust to growth shocks; benefits of maintaining growth to avoid fragility diminish when government effectiveness is very weak (less than -1).

### Policy implications and conclusions
- Persistence of fragility underscores both hurdles to exit and benefits of avoiding entry.
- Main policy implications:
  1. Near-fragile countries should adopt timely and appropriate counter-cyclical policies to preserve economic activity.
     - Enhanced fiscal stimulus is needed to prevent a sharp output contraction, but effect is weakened in countries with very weak government effectiveness.
     - International community should provide financing subject to appropriate governance safeguards and policy design.
     - In good times, countries should build and maintain external and fiscal buffers through prudent policies to enable counter-cyclical responses when needed.
  2. Improving institutions and enhancing political and social inclusion is critical for exit sustainability.
     - Protecting social spending (health and education) supports political and social inclusion, which with good policies fosters growth and resilience.
     - Fiscal capacity building (revenue mobilization and PFM) complements institutional development and state building.
     - Economic gains from exiting fragility (per SCM) can finance further institution, human capital and infrastructure building.
  3. Seize “pivotal moments” to embrace new approaches and reforms.
     - Pivotal moments create opportunities to reset citizen expectations and build trust; they are associated with stronger economic outcomes and policy/institutional improvements in sustained exit cases.
     - IFIs can provide technical assistance and capacity development during pivotal moments where political commitment to reform exists.
- Overall conclusions:
  - Three-pronged empirical approach (event studies, SCM, logit) highlights a nexus between growth/resilience and government effectiveness.
  - Avoiding growth collapses is crucial to prevent entries into fragility; improving government effectiveness is central to enable sustainable exits.
  - Complementary policies—macroeconomic buffers, social spending protection, institutional reforms, and international support with safeguards—are necessary to help countries avoid or exit the fragility trap.

*Source: IMF working paper chapter “6. Nexus of Growth and Government Effectiveness.”*

### REFERENCES

### wpiea2021133-print-pdf - REFERENCES

### Major referenced works and themes
- Methodological advances and econometric techniques:
  - Abadie, A., Diamond, A., and Hainmueller, J. 2010. Synthetic control methods for comparative case studies: estimating the effect of California’s tobacco control program. Journal of the American Statistical Association, Vol. 105, No. 490, pp. 493-505.
  - Agresti, A., 2007. An introduction to categorical data analysis, 2nd ed.
  - Bergtold, J. S., Yeager, E. A., and Featherstone, A. M. (2017). Inferences from logistic regression models in the presence of small samples, rare events, nonlinearity, and multicollinearity with observational data.
  - van der Ploeg, T., Austin, P. C., and Steyerberg, E. W. (2014). Modern modelling techniques are data hungry: a simulation study for predicting dichotomous endpoints.
- Fragility, state capacity, and conflict:
  - Acemoglu, Daron and James Robinson (2012, 2021) on institutions and inclusive states.
  - Besley, Timothy and Torsten Persson, 2010, 2011a, 2011b on state capacity and development.
  - Collier, Paul, 2007; Collier, Paul, 2020 on the poorest countries and transition programs.
  - Commission on State Fragility, Growth and Development, 2018, “Escaping the Fragility Trap.”
  - Multiple IMF papers and policy documents on fragile states, fiscal capacity, and IMF engagement (2010–2020).
  - World Bank, OECD, IGC, African Development Bank, and other institutional reports on fragility and conflict.
- Fiscal capacity, taxation, and public finance in fragile contexts:
  - Akitoby et al., 2019, 2020; Gaspar, Jaramillo, and Wingender, 2016a, 2016b; Gupta et al., 2005; Deléchat et al., 2018.
  - IMF Policy Papers and Working Papers on building fiscal capacity in fragile states.
- Economic costs and spillovers of conflict:
  - Mueller, H. (2013, 2016); Ncube, Jones, and Bicaba, 2014; Novta and Pugacheva, 2020; De Groot, 2010; Dirienzo and Das, 2017.
- Country and case studies:
  - Multiple Uganda country reports cited (IMF Country Report Nos. 07/29; 10/132; 13/215; 17/206; 19/125; 20/165).
  - Redifer et al., 2020 on Rwanda; Gelbard et al., 2015 on Sub-Saharan Africa fragile states.

### Appendix I — Definitions of Fragile States in IFIs and Key International Actors
- African Development Bank
  - Countries or situations with unique development challenges that have resulted from fragility and conflict including weak institutional capacities and poor governance, economic and geographic isolation, economic disruption, social disruption and insecurity.
- German Federal Ministry for Economic Cooperation and Development(BMZ)
  - Fragile statehood exists in situations where there is low level of government performance, where state institutions are weak or on the verge of collapse and where the state either fails to perform core roles or performs them wholly inadequately. The BMZ also refers to the CPIA (compiled by the World Bank).
- Country Indicators for Foreign Policy
  - Fragile states lack the functional authority to provide basic security within their borders, the institutional capacity to provide basic social needs for their populations, and/or the political legitimacy to effectively represent their citizens at home or abroad.
- Department for International Development (DFID)
  - DFID has used a broad definition (“Where the government cannot or will not deliver core functions to the majority of its people, including the poor.”) but also refers to a combination of the three widely accepted assessment frameworks: World Bank’s CPIA-indicators, the Fund for Peace’s Failed States Index (FSI) and the Uppsala Conflict Database.
- European Union
  - Fragility refers to weak or failing structures and to situations where the social contract is broken due to the state’s incapacity or unwillingness to deal with its basic functions, meets its obligations and responsibilities regarding service delivery, management of resources, rule of law, equitable access to power, security and safety of the populace and protection and promotion of citizens’ rights and freedoms.
- G7+
  - [A] state of fragility can be understood as a period of time during nationhood when sustainable socio-economic development requires greater emphasis on complementary peacebuilding and State-building activities such as building inclusive political settlements, security, justice, jobs, good management of resources, and accountable and fair service delivery.
- International Monetary Fund
  - Fragile states have characteristics that substantially impair their economic and social performance. These include weak governance, limited administrative capacity, chronic humanitarian crises, persistent social tensions, and often, violence or the legacy of armed conflict and civil war. In these countries the poor quality of policies, institutions and governance substantially impairs economic performance, the delivery of basic social services and the efficacy of donor assistance. Such states are least likely to achieve the MDGs. They also have considerable negative spillover effects on economic growth in neighboring countries.
- Organisation for Economic Co-Operation and Development (OECD)
  - Pockets of fragility may occur at a subnational level, making it hard to keep the fragile states terminology. The States of fragility report 2015 marks a change towards defining dimensions of fragility: violence, justice, institutions, economic foundations and resilience. Thus, the OECD breaks down the drivers of fragility for each country and reveals different patterns of vulnerability instead of trying to stringently define fragility.
- Swiss Agency for Development and Cooperation
  - A state or context is describe as fragile if a significant proportion of the population does not regard the state as the legitimate framework for the exercise of power, if the state does not or cannot exercise its monopoly of the legitimate use of force within its territory, and if the state is unable or unwilling to provide basic goods and services to a significant part of the population.
- United States Agency for International Development (USAID)
  - Fragile states refer to a broad range of failing, failed, and recovering states that are unable or unwilling to adequately assure the provision of security and basic services to significant portion of their populations and where the legitimacy of the governments is in question. USAID distinguishes between fragile states that are vulnerable from those that are already in
- World Bank
  - The World Bank defines fragile states according to their ranking in the Country Policy and Institutional Assessment that includes a set of 16 criteria grouped in four clusters: economic management, structural policies, policies for social inclusion and equity, and public sector management and institutions. The result is published every year in the “Harmonized List of Fragile Situations”. Fragile Situations include countries or territories with (i) a harmonized CPIA country rating of 3.2 or less, and/or (ii) the presence of a UN and/or regional peace-keeping or political/peace-building mission during the last three years.

### Appendix II — Data sources and indicator definitions
- Change in effective executive
  - Measures the number of times in a year that effective control of executive power changes hands. Such a change requires that the new executive be independent of his predecessor. This variable addresses one of drawbacks of the indicator related to major government changes as some cabinet changes may not entail change in executive power;
  - Source: Banks and Wilson (2019)'s Cross National Time Series Data Archive
- Conflict
  - Dummy variable. Number of casualties as a percent of the population in the previous year. If a country falls in the top 25th percentile, then it is calssified as having conflict
  - Source: Uppsala Conflict database
- CPIA
  - The CPIA rates countries against a set of 16 criteria grouped in four clusters: (i) economic management; (ii) structural policies; (iii) policies for social inclusion and equity; and (iv) public sector management and institutions. The criteria are focused on balancing the capture of the key factors that foster growth and poverty reduction, with the need to avoid undue burden on the assessment process.
  - Source: World Bank
- Education Expenditure (% of GDP)
  - Total expenditure on educatoin as a share of GDP
  - Source: WDI
- Executive constraints
  - The extent of institutionalized constraints on the decisionmaking powers of chief executives, whether individuals or collectivities. A seven-category scale is used, which includes unlimited authority; slight to moderate limitation on executive authority; substantial limitations on executive authority; executive parity or subordination.
  - Source: Polity IV
- Financial Crisis
  - Dummy variable. Takes the value of “one” when at least one of three crises occurs: currency, banking, or sovereign debt
  - Source: Laeven and Valencia, 2018
- GDP per capita growth
  - Gross domestic product, current prices, national currency, percent change
  - Source: WEO
- Government Effectiveness
  - Government effectiveness captures perceptions of the quality of public services, the quality of the civil service and the degree of its independence from political pressures, the quality of policy formulation and implementation, and the credibility of the government's commitment to such policies
  - Source: Worldwide Governance Indicators
- Government expenditure (% of GDP)
  - Total government expenditure as a share of GDP
  - Source: WDI
- HDI
  - An index of life expectancy, education, and per capita income indicators, which are used to rank countries into four tiers of human development
  - Source: United Nations
- Health Expenditure (% of GDP)
  - Total expenditure on health as a share of GDP
  - Source: WDI
- Infant Mortality
  - Number of death per 1000 live births
  - Source: WDI
- Inflation
  - Consumer Prices, end of period, percent change (Percent , Units)
  - Source: WEO
- Life expectancy
  - Life Expectancy at birth (in years)
  - Source: WDI
- Military Expenditure (% of GDP)
  - Total expenditure on military as a share of GDP
  - Source: WDI
- Pivotal Moments (2 definitions)
  - Pivotal moments---crises (have “one” when at least one of four crises occurs: currency, banking, sovereign debt, and debt restructuring), and change in executive power. First, create dummies. Then, plot three cases: crises, change in chief executives, and sum of both.
  - Source: Collier, 2020
- Political Competition
  - Political Competition combines two concepts, first, the degree of institiutionalization or regulation of political competition and second, the extent of government restriction on political competition
  - Source: Polity IV
- Political Stability
  - Political Stability and Absence of Violence/Terrorism measures perceptions of the likelihood of political instability and/or politicallymotivated violence, including terrorism
  - Source: Worldwide Governance Indicators
- Regulatory Quality
  - Regulatory quality captures perceptions of the ability of the government to formulate and implement sound policies and regulations that permit and promote private sector development
  - Source: Worldwide Governance Indicators
- Tax Revenue as a percent of GDP
  - General government taxes, percent of fiscal year GDP (Percent of GDP, Units)
  - Source: WEO
- Total Investment (Share of total spending)
  - Domestically financed capital expenditure as a share of total spending
  - Source: WEO

*Source: wpiea2021133-print-pdf - REFERENCES*

### Appendix III. Regression Tables:

### Appendix III. Regression Tables

### Entry Events, by Logit Model
Dependent variable: Entry vs Counterfactual

- GDP per capita growth, year T-3 to T-1, average
  - Model (1): -47.2975*** (14.2282)
  - Model (2): -42.6886*** (13.3086)
  - Model (3): -43.8960*** (13.3384)
  - Model (4): -39.4549** (15.9038)
  - Model (5): -38.7512*** (13.2831)
  - Model (6): -38.2088*** (13.2134)
  - Model (7): -38.4797** (16.3812)

- Government effectiveness, year T-1
  - Model (1): -1.9832* (1.1207)
  - Model (2): -1.6427* (0.9413)
  - Model (3): -1.4550 (0.9646)
  - Model (4): -1.6672 (1.1036)
  - Model (5): -1.7839* (1.0237)
  - Model (6): -1.9712* (1.0664)
  - Model (7): -1.5852 (1.1120)
  - Model (8): -1.3650* (0.8006)
  - Model (9): -1.9026** (0.9634)
  - Model (10): -1.3014 (0.9242)
  - Model (11): -1.8246** (0.9120)

- Life expectancy at birth, year T-1
  - Model (1): -0.0375 (0.0400)
  - Model (2): -0.0159 (0.0321)
  - Model (3): -0.0278 (0.0356)
  - Model (4): -0.0342 (0.0344)
  - Model (5): -0.0276 (0.0329)

- Total Revenue as a % of GDP
  - Model (1): 0.0055 (0.0231)

- Inflation
  - Model (2): -0.0215 (0.0348)
  - Model (3): -0.0102 (0.0302)

- Executive Constraints, year T-1
  - Model (2): -0.1302 (0.1951)
  - Model (3): -0.0530 (0.1727)

- WEO: Terms of trade, goods, US Dollars, percent change (Percent , Units)
  - Model (2): -0.0018 (0.0132)
  - Model (3): -0.0059 (0.0122)

- L&V: Dummy that takes the value 1 if the country had any kind of financial crisis.
  - Model (2): 0.4664 (1.1650)
  - Model (3): 0.6928 (1.1334)

- Polity: Political Competition ; Political Competition
  - Model (2): -0.0153 (0.1779)

- Constant (selected models)
  - Model (1): -2.4847 (2.8094)
  - Model (2): -0.2051 (0.9526)
  - Model (3): 0.2789 (0.8738)
  - Model (4): -0.0164 (1.4045)
  - Model (5): -0.3611 (0.8218)
  - Model (6): -0.5841 (0.8926)
  - Model (7): -0.4561 (1.5528)
  - Model (8): 0.1647 (2.2436)
  - Model (9): 0.2019 (2.5287)
  - Model (10): 1.1063 (2.4383)
  - Model (11): 0.2528 (2.3135)

- Observations by model
  - Models (1)–(11): 67, 67, 68, 53, 61, 63, 52, 69, 55, 62, 65 (presented as "Observations6767685361635269556265")

- Significance notation
  - *** p<0.01, ** p<0.05, * p<0.1
  - Standard errors in parentheses

Additional models (12)–(22) (dependent variable: Entry vs Counterfactual) — selected coefficients and statistics:

- Government effectiveness, year T-1
  - Model (12): -1.6562* (0.9634)

- Life expectancy at birth, year T-1
  - Models (12)–(16): -0.0375 (0.0372); -0.0357 (0.0298); -0.0361 (0.0292); -0.0325 (0.0275); -0.0401 (0.0309)

- Inflation (Models (12)–(18))
  - Coefficients: -0.0160 (0.0257); -0.0394 (0.0335); -0.0148 (0.0286); -0.0228 (0.0264); -0.0443 (0.0391); 0.0002 (0.0310); 0.0002 (0.0310)

- Executive Constraints, year T-1 (Models (12)–(17))
  - Coefficients: -0.0083 (0.1491); 0.0323 (0.1614); -0.0498 (0.1508); -0.0498 (0.1508); -0.0552 (0.1610); -0.0552 (0.1610)

- WEO: Terms of trade, goods, US Dollars, percent change (Percent , Units)
  - Models (12)–(16): -0.0076 (0.0116); -0.0381 (0.0257); -0.0448 (0.0275); -0.0448 (0.0275); -0.0483* (0.0284)

- L&V: Dummy for financial crisis
  - Models (12)–(17): -0.3749 (1.0359); -1.1052 (1.3046); -1.1052 (1.3046); -1.0720 (1.1825); -1.0720 (1.1825); -1.1903 (1.1912)

- Polity: Political Competition
  - Models (18)–(20): 0.1014 (0.1534); 0.0906 (0.1302); 0.0693 (0.1350)

- Constant (Models (12)–(22)) — selected entries
  - Model (12): 0.1138 (2.7340)
  - Model (13): 1.7295 (1.8424)
  - Model (14): 2.2160 (1.7556)
  - Model (15): 1.8035 (1.6770)
  - Model (16): 1.4847 (2.0613)
  - Model (17): -0.3255 (0.7063)
  - Model (18): -0.2794 (0.6544)
  - Model (19): -0.2794 (0.6544)
  - Model (20): -0.3281 (0.6991)
  - Model (21): -0.3281 (0.6991)
  - Model (22): -0.9792 (0.8629)

- Observations by model (12)–(22)
  - Models (12)–(22): 54, 67, 73, 76, 66, 66, 26, 66, 66, 61, 61 (presented as "Observations5467737666626666616161")

- Significance notation
  - *** p<0.01, ** p<0.05, * p<0.1
  - Standard errors in parentheses

---

### Exit Events, by Logit Model
Dependent variable: Exit vs Counterfactual

- GDP per capita growth, year T-3 to T-1, average
  - Model (1): 13.1823* (7.1244)
  - Model (2): 13.5508* (7.1947)
  - Model (3): 15.9313* (8.9134)
  - Model (4): 19.6929** (8.7387)
  - Model (5): 16.9690** (8.6185)
  - Model (6): 16.1826* (8.9142)

- Government effectiveness, year T-1
  - Model (1): 2.6118** (1.0294)
  - Model (2): 2.6675*** (0.9313)
  - Model (3): 3.8820*** (1.4297)
  - Model (4): 3.3446*** (1.1734)
  - Model (5): 3.0562*** (1.1262)
  - Model (6): 3.4325*** (1.3085)
  - Model (7): 2.3575** (0.9317)
  - Model (8): 3.1956** (1.3020)
  - Model (9): 2.8740*** (1.0793)
  - Model (10): 2.6027** (1.0573)
  - Model (11): 3.0018** (1.2517)

- Life expectancy at birth, year T-1
  - Models (1)–(6): 0.0050 (0.0419); 0.0151 (0.0419); 0.0480 (0.0535); 0.0205 (0.0448); 0.0339 (0.0456); 0.0461 (0.0512)

- Inflation
  - Models (2) and (3): -0.0187 (0.0567); -0.0085 (0.0549)

- Executive Constraints, year T-1
  - Models (2) and (3): -0.3298 (0.2840); -0.2619 (0.2625)

- WEO: Terms of trade, goods, US Dollars, percent change (Percent , Units)
  - Models (2) and (3): -0.0351 (0.0220); -0.0256 (0.0185)

- L&V: Dummy that takes the value 1 if the country had any kind of financial crisis
  - Model (7): -- (no coefficient shown)

- Polity: Political Competition
  - Models (2) and (3): -0.1478 (0.2170); -0.0643 (0.1932)

- Constant (selected models)
  - Model (1): 1.1237 (3.0051)
  - Model (2): 1.5843* (0.8631)
  - Model (3): 4.2305** (1.9454)
  - Model (4): 2.0092** (1.0061)
  - Model (5): 1.9342* (0.9882)
  - Model (6): 3.4610** (1.7580)
  - Model (7): 0.7146 (2.9762)
  - Model (8): 0.9603 (3.7116)
  - Model (9): 0.8813 (3.1556)
  - Model (10): -0.0011 (3.1467)
  - Model (11): 0.2901 (3.6381)

- Observations by model
  - Models (1)–(11): 59, 59, 40, 53, 50, 41, 59, 40, 53, 50, 41 (presented as "Observations5959405350415940535041")

- Significance notation
  - *** p<0.01, ** p<0.05, * p<0.1
  - Standard errors in parentheses

---

### Appendix VIII. Case Studies

### 1. Rwanda — Findings and policy-relevant factors
- Political commitment, accountability, reform ownership, and international partner support helped sustain reforms after regaining political stability.
- Pre-exit conditions:
  - CPIA was only slightly above 2 (mid-1990s).
- Institutional rebuilding focused on revenue administration, budget, and banking systems.
- Fiscal and macro outcomes:
  - Rwanda’s CPIA score increased from only 2.2 at mid-1990s to consistently above 3.5 after the exit.
  - Rwanda’s tax-to-GDP ratio increased from 9.7 percent in 2000 to over 13 percent on average during 2010-13.
  - Priority spending increased from 4 percent of GDP in 1999 to 12-14 percent during 2008-12.
  - Poverty rate declined from 75 percent (on average) in 1992-2000 to around 59 percent in 2010-16.
- Role of external support:
  - Fund-supported programs and technical support aimed at enhancing revenue efforts and boosting international reserves.
  - HIPC and MDRI aided debt reduction, creating fiscal space.
  - IFIs and bilateral donors provided technical support on revenue mobilization, PFM, and monetary and exchange rate management.
- Key reform elements highlighted:
  - Tax administration: utilize information management systems and improve tax compliance.
  - Tax policy: raise rates of several indirect taxes.
  - PFM: reinstitute the budget process with parliament adopting annual budget laws since 1998 and rebuilding PFM systems by the mid-2000s.
- Concluding point:
  - Reform efforts amid political stability, backed by international support, improved resilience, governance, institutions, and social inclusion enabling sustained exit from fragility.

### 2. Uganda — Findings and policy-relevant factors
- Context and timing:
  - Uganda exited from fragility around the middle of the 2000s as conflicts subsided; peace talks started in 2006.
- Macroeconomic and fiscal performance:
  - Strong macroeconomic performance supported by prudent macroeconomic policies, a sound banking sector, and substantial donor assistance.
  - CPIA stayed well above the threshold (slightly less than 4, against the threshold 3.2).
  - Tax revenue increased from slightly below 9½ percent of GDP on average before the start of the peace talks to over 11 percent on average during the second half of the 2010s.
  - Poverty: share below the poverty line declined from 65 percent in 2002 to 36 percent in 2012.
- Key reforms:
  - Revenue-side reforms: eliminate tax exemptions, make income tax more progressive, reform excise taxes, improve tax administration (e.g., high net worth individuals unit, strengthen medium-sized taxpayer office), and widely use electronic tax services.
  - PFM reforms: strengthen spending controls, including setting up a Treasury Single Account.
- External support and programs:
  - IMF policy advice and three Policy Support Instrument programs (PSI); Fund and donors provided technical support (monetary policy framework, financial system reform, revenue mobilization, PFM).
  - Note: some progress reversed after PSI programs ended in 2017 (e.g., tax exemptions increased; governance indicators such as the control of corruption index worsened; gradual withdrawal of on-budget donor support).
- Concluding point:
  - Improved political stability provided the basis for implementing reforms that strengthened the macroeconomic policy framework and built resilience.

*Source: wpiea2021133-print-pdf — Appendix III. Regression Tables; Appendix VIII. Case Studies*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021133-print-pdf.pdf_