## _wp04184

## Source details

**Canonical URL:** [_wp04184](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04184.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04184.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04184.pdf.json)

---

### Overview and scope
- Sample: 197 IMF programs approved between 1992 and 2002.
- Comparative focus: interrupted versus completed IMF-supported programs; use of MONA implementation measures and ICRG institutional/political indicators.
- Key implementation measures:
  - Binary irreversible-interruption index (major and irreversible interruption defined as either the last scheduled program review not completed, or all scheduled reviews completed but the subsequent annual arrangement not approved).
  - Ratio of disbursements to commitments (continuous disbursement share).
- Key macroeconomic outcomes analyzed: inflation, economic growth, fiscal balance-to-GDP ratio, current account-to-GDP ratio.
- Main empirical approaches: correlations, first-stage regressions of implementation measures, and outcome regressions with IV methods (2SLS, 3SLS) to account for endogeneity.

### Measures, stylized facts, and descriptive results
- Implementation and institutional descriptive statistics:
  - 41 percent of all programs (including precautionary arrangements) experienced an irreversible interruption; IMMA (2003) reports about 44 percent.
  - Correlation coefficient between program interruptions and the disbursement share is -0.7.
  - When precautionary arrangements are excluded, average disbursement share is approximately 75 percent, compared with 71 percent in the IMMA (2003) sample.
  - Programs with fewer interruptions tend to have higher disbursement rates.
  - Improvements in ICRG indicators are generally positively correlated with higher disbursement shares and fewer interruptions.

- Patterns across interrupted vs. uninterrupted programs (Figures referenced in source):
  - Interrupted programs: less progress in investment climate and bureaucracy quality; intensified internal conflict relative to the preprogram year.
  - Successful implementation: greater initial military influence followed by significant reduction; interruptions show sharp increases in military involvement in years three and four after approval.
  - Inflation: countries completing programs more successful in reducing inflation (mean changes between period T - 1 and five horizons).
  - Fiscal balance: uninterrupted programs show sharp improvement in year one then gradual deterioration; interrupted programs show modest initial improvement then catch up.
  - Current account: uninterrupted programs show improvement; interrupted programs show slight immediate improvement then steady deterioration.

### Instrumental variables and identification
- Instruments listed and their intended roles:
  - IMF quota (log).
  - Number of months spent in IMF-supported programs since 1980.
  - Amount approved as a fraction of IMF quota.
  - IMF effort per program year (including six months prior to program approval), computed from Budget Reporting System staff hours and estimated average staff salaries.
- IV rationale: instruments must be correlated with implementation measures (disbursement share and interruption index) and plausibly exogenous to direct determinants of inflation, growth, fiscal balance, and current account outcomes.
- Overidentifying restrictions: tests could not reject the null for either implementation measure.

### Econometric framework and estimation strategy
- Reduced-form AR/mean-reversion representation for each macroeconomic outcome:
  - ΔM_Ti = γ0 + γ1 ΔM_{T-1,i} + γ2 M_{T-1,i} + γ3 IMPL_i + γ4 INST_{T-1,i} + γ5 ΔINST_{T,i} + ε_{Ti}
- Program implementation model (first-stage):
  - IMPL_i = θ0 + θ1 INITIAL_i + θ2 FUND_i + θ3 INST_{T-1,i} + ε_i
  - INITIAL includes preprogram real GDP per capita, inflation, GDP growth rate, current account balance, fiscal balance.
  - FUND includes program-specific variables related to IMF financial/human resource effort and program size.
- Treatment of institutions and politics:
  - ICRG indices treated as exogenous and mean-reverting; both lagged levels and first differences included to capture temporary deviations.
  - Parsimonious "testing down" approach: start with unrestricted model containing all ICRG indices and drop regressors with low t-statistics; adjusted R2 used for model selection.
- Endogeneity addressed via:
  - Two-stage least squares (2SLS), three-stage least squares (3SLS); probit and tobit used for interruption dummy and disbursement share when appropriate.
  - 3SLS uses 2SLS estimates to obtain the variance-covariance matrix and achieves greater efficiency.
- Choice of AR/mean-reversion over full VECM: VECM considered and results robust; AR representation used to conserve degrees of freedom and simplify presentation.

### First-stage determinants of implementation (key coefficients and effects)
- Disbursement share (continuous implementation measure):
  - Institutional/political effects:
    - A one-point increase in either ethnic tension rating (reduction in tension) or government stability rating raises disbursements by about 8 percent (ceteris paribus).
    - A one-point improvement in the internal conflict rating (reduction in conflict) lowers disbursements by just over 6 percent (internal conflict coefficient negative, significant at the 1 percent level).
    - Law and order rating coefficient negative and significant at the 10 percent level (improvements associated with lower disbursements in the year before approval).
    - Greater initial military involvement associated with lower disbursements (ICRG index coefficient positive and significant at the 10 percent level).
  - Program history and size:
    - Every additional month spent in IMF-supported programs translates into 0.2 percent more funds disbursed.
    - Larger programs (amount approved relative to quota) associated with higher disbursement shares.
  - IMF effort measures (staff hours, dollar cost): only marginally important.
  - Country’s IMF quota coefficient negative, suggesting weaker implementation in countries with larger Fund quotas.

- Irreversible-interruption index (binary implementation measure):
  - Initial economic conditions mostly insignificant; exception: lagged level of growth rate (faster preprogram growth → less likely irreversible interruption).
  - Corruption:
    - A one-point improvement in the ICRG corruption index coincides with a 35.8 percent better chance of having no program interruption (all other determinants held constant).
  - Internal conflict:
    - Initial level negative and significant; an improvement in this rating by one point is associated with a 21.7 percent higher chance of an irreversible interruption (political violence coefficient reported with this sign in source).
  - Relationship-with-Fund variables: expected signs but generally insignificant except months in program mode.

- Overall interpretation from first-stage results:
  - Program implementation largely determined by domestic political economy and institutions.
  - Fund-controlled variables have only marginal impact on implementation outcomes.

### Main empirical findings by macroeconomic outcome
- Inflation:
  - High persistence: lagged inflation coefficients highly significant.
  - Speed of reversal:
    - For the average program, about three quarters of any deviation from “normal” inflation after a program is approved is reversed within a year.
    - Deviations are erased almost completely in three years.
  - Program implementation effects:
    - After accounting for endogeneity, disbursement share coefficients negative and declining in absolute value over three years; only the first-year result statistically significant.
    - Lack of program interruptions correlated with greater price stability in year following approval, followed by slightly higher inflation over longer horizons.
  - Institutional effects:
    - Better government bureaucracy at program start and improvements subsequently → lower inflation (strongest in year one).
    - Better law and order and improvements → lower inflation, though better law and order before start and improvements during program associated with slightly higher inflation at T + 2.
  - Political factors:
    - Lower inflation associated with increased political violence (change in ICRG Internal Conflict rating) in first two years; interpreted as demand-side austerity lowering inflation but raising unrest.
  - Robustness: excluding observations with annual change in inflation greater than 50 percent (sample reduced by approximately 30 percent) changes some significance patterns (ethnic tensions remain important; internal conflicts become insignificant; government stability significant).

- Economic growth:
  - High serial correlation and mean reversion:
    - Approximately 90 percent of any deviation in growth rates from the country’s “normal” growth pattern is made up within three years.
    - Largest adjustment (83-84 percent) within one year.
  - Program implementation effects:
    - OLS suggests positive and significant association with disbursement share.
    - 2SLS and 3SLS estimates not significantly different from zero (no statistically significant association detected within three years).
    - Implementation effects on growth may operate beyond the three-year horizon.
  - Institutional effects:
    - Investment Profile: a one-point increase associated with roughly a 1 percent increase in the growth rate (robust across horizons and estimation techniques).
    - Improvements in government stability positively impact growth, especially in the first year.
  - Robustness: results hold when omitting outliers (countries with growth changes > 10 percent per annum).

- Fiscal balance (percent of GDP):
  - Persistence and mean-reversion:
    - Improvements persist for two years then reverse.
    - Approximately 45 percent of any deviation of the fiscal balance from long-run average is offset within a year.
  - Program implementation effects:
    - OLS: implementation associated with fiscal improvements (first two years).
    - IV (2SLS, 3SLS): implementation associated with larger fiscal deficits; IV estimates of disbursement share coefficient two years after approval are significant.
    - Quantification: fiscal deficits in countries with completed programs are about 3 percent larger than in countries whose programs were interrupted (source statement).
    - Interpretation: better implementation brings additional financing, enabling more gradual adjustment and larger fiscal deficits initially.
  - Institutional effects:
    - Initial level of corruption important but anomalous: lower corruption associated with weaker fiscal outcomes over time in the sample.
    - Improvements in Democratic Accountability associated with larger deficits (possible electoral-cycle effects).
    - Declines in ethnic tension contemporaneously correlated with improved fiscal balances.
    - Less military involvement preprogram and declines in external conflict associated with lower fiscal deficits.

- Current account (percent of GDP):
  - Shocks are longer lived; larger permanent components:
    - Only about 10 percent of any deviation from a country’s “normal” current account-to-GDP ratio is made up in one year.
  - Program implementation effects:
    - Better implementation associated with deterioration of the current account for about two years, followed by sharp improvement in third year.
    - Quantification: disbursement of 100 percent of committed funds is accompanied by an 8 percent deterioration of the current account in the first year (relative to the preprogram year), followed by a 2 percent improvement in the third year.
  - Institutional effects:
    - Investment Profile (initial level and change) highly significant with negative signs—better investment climate associated with larger current account deterioration during program period.
  - Political effects:
    - Change in ICRG External Conflict index negative and highly significant for T + 2: a one-point increase correlated with a 1.2 percent deterioration of the current account over three years.
    - Preprogram government stability and less military involvement associated with better current account performance.
  - Robustness: results robust to choice of implementation measure.

### Relative contributions and dynamics (empirical decomposition)
- Inflation:
  - Program implementation drives inflation dynamics in the first and second year.
  - Impact of implementation roughly 3½ and 2 times greater than autoregressive/mean-reversion dynamics for horizons T and T + 1, respectively.
  - Institutional and political factors push inflation upward and are especially important over two-year horizon (twice the size of mean-reversion terms).
- Growth:
  - Both program implementation and institutional/political factors matter over one- and three-year horizons; institutional/political conditions have quantitatively greater impact.
  - Program effects are growth-stimulating at all horizons; institutional/political factors reduce growth in short run but raise it in long run.
- Fiscal balance:
  - Institutional factors and program implementation influence fiscal balance in similar magnitudes but in opposite directions.
  - Institutional factors especially important over one- and three-year horizons; program implementation very important over two-year horizon.
- Current account:
  - Program implementation is major influence in first year—impact more than seven times greater than autoregressive/mean-reversion factors.
  - Implementation tends to worsen the current account short run and improve it long run.
  - Institutional factors' relative importance increases with horizon length.

### Conclusions and policy implications
- Four main conclusions:
  1. Institutional quality and domestic political environment matter materially for macroeconomic outcomes in program countries up to three years; institutional improvements have both direct effects and indirect effects via improved implementation.
  2. Institutional and political environment are quantitatively important determinants of program implementation (higher disbursements and fewer interruptions where ethnic tensions low, governments stable, corruption lower, and military less involved).
  3. Program implementation varies with duration of financial engagement and quota-adjusted program size: more months in prior IMF programs and larger programs (relative to quota) associated with better implementation.
  4. After accounting for institutional impacts, program implementation independently affects macroeconomic outcomes up to two years: better implementation associated with lower inflation and initially weaker but ultimately stronger external and fiscal outcomes; no statistically significant impact on economic growth within three years.

- Policy recommendations and operational suggestions for the IMF:
  - Research agenda:
    - Investigate whether successful program implementation affects growth beyond the three-year horizon.
    - Identify structural reforms that generate quicker growth payoffs for program design and implementation.
  - Operational guidance:
    - Monitor quantitative institutional and political indicators (e.g., ICRG ratings of ethnic tension, government accountability, investment climate) during program design and implementation to assess prospects for growth and implementation.
    - Use high-frequency information on determinants of investment profile (contract viability, expropriation risk, repatriation ease, payment delays) as implementation feedback.
    - Incorporate quantitative political and institutional analysis into surveillance and program design to provide early warning of declining implementation prospects; consider institutionalizing measures to secure endorsement from major political forces to improve ownership.
  - Modeling recommendation:
    - Move beyond treating institutions as exogenous and mean-reverting; model institutional evolution endogenously to assess feedback from program implementation to institutional and political dynamics.

*Source: _wp04184 (selected sections and Box 1 text from the provided IMF document).*

### 1.         Success of Program Implementation by Type of Arrangement...................................25

### 1.         Success of Program Implementation by Type of Arrangement...................................25

### Overview of content sections
- 1.         Success of Program Implementation by Type of Arrangement...................................25
- 2.         Correlations Between Implementation Indices ...........................................................26
- 3. Correlations of Implementation Indices with ICRG Risk Ratings at  
            Different            Horizons  .......................................................................................................27
- 4. First Stage Regressions of Implementation Measures on Initial Economic 
            Conditions, ICRG Indicators, and Instruments............................................................28
- 5. Regressions of the Change in Inflation on the Share of Committed Funds  
            Disbursed  .....................................................................................................................29
- 6.         Regressions of the Change in Inflation of the Irreversible Interruption Index............30
- 7.         Regressions of Economc Growth on the Share of Committed Funds Disbursed  ........31
- 8.         Regressions of Economic Growth on the Irreversible Interruption Index...................32
- 9. Regressions of the Fiscal Balance-to-GDP Ratio on the Share of Committed  
            Funds            Disbursed...........................................................................................................33
- 10. Regressions of the Fiscal Balance-to-GDP Ratio on the Irreversible Interruption  
            Index  ............................................................................................................................34
- 11. Regressions of the Current Account-to-GDP Ratio on the Share of Committed  
            Funds            Disbursed...........................................................................................................35
- 12. Regressions of the Current Account-to-GDP Ratio on the Irreversible Interruption 
            Index  ............................................................................................................................36

### Figures included in this content unit
- Figure 1.  The Quality of Institutions in Interrupted and Completed Programs ..........................22
- Figure 2.  Macroeconomic Performance in Interrupted and Completed Programs......................24

### Key methodological focus (as indicated by section titles)
- Use of implementation indices and their correlations (Sections 2 and 3).
- First-stage regressions linking implementation measures to initial conditions, ICRG indicators, and instruments (Section 4).
- Outcome regressions examining the relationship between implementation (share of committed funds disbursed; irreversible interruption index) and macroeconomic variables:
  - Change in inflation (Sections 5 and 6).
  - Economic growth (Sections 7 and 8).
  - Fiscal balance-to-GDP ratio (Sections 9 and 10).
  - Current account-to-GDP ratio (Sections 11 and 12).

### Implied analytical structure
- Comparative analysis of interrupted versus completed programs (Figures 1 and 2).
- Statistical inference through correlations and regression analysis across multiple macroeconomic outcomes and implementation measures.
- Consideration of institutional quality (Figure 1) and program implementation success by arrangement type (Section 1).

*Source: _wp04184 - 1.         Success of Program Implementation by Type of Arrangement...................................25*

### Box 1.  List of Instrumental Variables  ................................................................................

### _wp04184 - Box 1.  List of Instrumental Variables

### Introduction
- The quality of a country’s institutions—broadly defined to include the formal and informal rules of economic and political interactions—is a key determinant of sustainable economic progress.
- Weak institutions are behind many development failures (IMF, 2003; Rodrik, Subramanian, and Trebbi, 2002; North, 1997).
- Institutional failings also affect the implementation and success of IMF-supported programs.
- Most prior empirical evaluations of IMF-supported programs:
  - Ignore the effect on macroeconomic performance of variation in institutions across countries or over time.
  - Are generally inconclusive about effects on economic growth, though inflation, the balance of payments, and the public finances seem to improve in adopting countries.
- Program implementation matters for program success; many IMF-supported programs experience major interruptions (IMMA, 2003; Mecagni, 1999).
- Recent literature investigates incidence and determinants of program implementation and macroeconomic consequences (IMMA, Joyce, 2003; Dreher, 2004).
- This paper measures the effect of variation in institutional quality on macroeconomic performance in countries implementing IMF-supported programs, using:
  - Program implementation measures from the Fund’s Monitoring of Fund Arrangements (MONA) database.
  - Institutional and political measures from the International Country Risk Guide (ICRG).

### Measuring Program Implementation and Institutional Development
- Two complementary measures of program implementation:
  - A binary irreversible-interruption index indicating whether a program experienced a major and irreversible interruption (defined as either the last scheduled program review not completed, or all scheduled reviews completed but the subsequent annual arrangement not approved). The index was updated using MONA.
  - The ratio of disbursements to commitments (continuous), indicating the share of available IMF credit actually drawn; captures actual program duration and the extent to which IMF financial commitments were fulfilled.
- ICRG political risk indicators used to capture institutional and political setting; 12 components divided into two groups:
  - Basic institutional quality, protection of property rights, and contract enforcement: Investment Profile, Corruption in Government, Law and Order, Quality of the Bureaucracy.
  - Political outcomes: Government Stability, Socioeconomic Conditions, Internal Conflict, External Conflict, Military in Politics, Religious Tensions, Ethnic Tensions, Democratic Accountability.
- Descriptive statistics and stylized facts:
  - Sample: 197 IMF programs approved between 1992 and 2002.
  - 41 percent of all programs (including precautionary arrangements) experienced an irreversible interruption, compared with about 44 percent reported in IMMA (2003).
  - Correlation coefficient between program interruptions and the disbursement share is -0.7 (Table 2).
  - When precautionary arrangements are excluded, the average disbursement share is approximately 75 percent, compared with 71 percent in the IMMA (2003) sample.
  - Programs with fewer interruptions tend to have higher disbursement rates.
  - Improvements in ICRG indicators are generally positively correlated with better program implementation (higher disbursement shares and fewer interruptions).
  - Greater government stability and a stronger investment profile in the year preceding program approval are associated with fewer interruptions (Table 3).
  - Improvements in the investment profile over approximately two years after program start lead to significantly fewer interruptions.
  - Lower corruption (higher ICRG score) and improvements in socioeconomic conditions in the year after program approval are associated with better implementation as measured by the quantitative implementation index.
- Patterns across interrupted vs. uninterrupted programs:
  - Interrupted programs show less progress in investment climate and bureaucracy quality, and intensified internal conflict during the program relative to the last preprogram year (Figure 1).
  - Successful implementation associated with greater initial military influence followed by significant reduction; interruptions accompanied by sharp increases in military involvement in the third and fourth year after approval.
  - Countries completing programs are more successful in reducing inflation than countries with interrupted programs (mean changes between period T - 1 and five horizons; Figure 2).
  - Uninterrupted programs show sharp fiscal balance improvements in the first year, then gradual deterioration; interrupted programs show modest initial fiscal improvements then catch up.
  - External current account balance improves for uninterrupted programs; interrupted programs show slight improvement in the year immediately following approval, then steady deterioration.
- Correlation analysis and mean comparisons mask wide variability due to differing initial conditions and external/internal shocks; necessitates more rigorous econometric methods.

### Econometric Methodology and Results
- Data approach:
  - Pooled dataset treating each program as an independent observation.
  - Framework accounts for short-run autoregressive, mean-reverting nature of macroeconomic variables.
  - Focus on impact of program implementation on growth, inflation, fiscal balance, and external balance while controlling for country-specific institutional and political effects.
  - Does not address sample selection bias between countries that choose to participate in IMF programs and those that do not.
- Model setup:
  - Let M_Ti denote a macroeconomic variable observed at time T in country i (country-program identifier).
  - General reduced-form representation:
    - ΔM_Ti = f(X_{T-1,i}, IMPL_i, INST_{T-1,i}, ΔINST_{T,i})
    - Where X_{T-1} is vector of non-institutional forcing variables at time T - 1 including a random term in time T; IMPL is program implementation; INST_{T-1} is initial institutional/political conditions; ΔINST_T is contemporaneous institutional changes.
  - Example AR(2) representation for M with autoregressive and mean-reversion form resulting in equation (4):
    - ΔM_Ti = γ0 + γ1 ΔM_{T-1,i} + γ2 M_{T-1,i} + γ3 IMPL_i + γ4 INST_{T-1,i} + γ5 ΔINST_{T,i} + ε_{Ti}
    - Interpretation: γ2 is a partial adjustment coefficient showing percentage of deviation from long-run equilibrium covered each year; small negative values consistent with highly persistent M.
  - Institutional and political developments during program treated as exogenous and mean-reverting; both first differences and lagged levels of institutional variables included to reflect temporary deviations from long-run levels.
  - Little correlation in the data between program implementation and institutional development (bottom panel of Table 3, first two columns).
- System of equations estimated (one for each macroeconomic outcome: inflation, growth, fiscal balance, current account balance):
  - Each outcome equation in system (5) has form:
    - ΔOutcome_Ti = γ0 + γ1 ΔOutcome_{T-1,i} + γ2 Outcome_{T-1,i} + γ3 IMPL_i + γ4 INST_{T-1,i} + γ5 ΔINST_{T,i} + ε_{Ti}
  - Monetary and fiscal policies are kept in the background; program implementation serves as a proxy for impact of macroeconomic policies.
- Program implementation model:
  - IMPL_i = θ0 + θ1 INITIAL_i + θ2 FUND_i + θ3 INST_{T-1,i} + ε_i
  - INITIAL: preprogram values of real GDP per capita, inflation, GDP growth rate, current account balance, fiscal balance.
  - FUND: program-specific variables related to IMF financial and human resource effort or nature of country–Fund relationship.
- On choice between VECM and AR representation:
  - A VECM was considered and confirmed existence of long-run relationships among some macro variables.
  - Marginal benefit of full VECM for the study’s focus was small; testing-down yields identical institutional/political factor specifications; coefficients and significance change marginally.
  - To conserve degrees of freedom and simplify presentation, the AR/mean-reversion representation is used; VECM results available from authors.
- Endogeneity concerns and instrumental variables:
  - Shocks to macroeconomic outcomes likely impact program implementation implying Corr(M_{Ti}, ε_{IMPL}) ≠ 0, causing OLS bias.
  - Example: a drought can lower tax revenue and raise the budget deficit; Fund may issue waivers and continue disbursements, generating correlation between implementation and errors.
  - Two IV techniques employed:
    - Two-stage least squares (2SLS): first regress implementation on exogenous variables and instruments correlated with implementation but uncorrelated with error terms in outcome equations; second stage replaces actual implementation with predicted values in outcome regressions and estimates by OLS.
    - (A second related IV technique is indicated though not fully detailed in the excerpt.)
- Main empirical findings (summary statements from text):
  - When endogeneity of program implementation is properly accounted for, institutions and program implementation both matter for macroeconomic performance.
  - Responses of macroeconomic variables to programs are often nonmonotonic:
    - Better-implemented programs are associated with lower inflation rates.
    - Fiscal and external current account balances typically deteriorate for two years after program approval before they turn around.
  - No statistically significant association detected between program implementation and economic growth at any time during the three years following program approval (consistent with previous work).

*Source: _wp04184 - Box 1.  List of Instrumental Variables (excerpt).*

### 0.40 for inflation, growth, fiscal balance, and current account balance, respectively).

### _wp04184 - 0.40 for inflation, growth, fiscal balance, and current account balance, respectively).

### B. Model Specification
- Parsimonious "testing down" approach:
  - Start with an unrestricted model including all ICRG indices; simplify by systematically dropping regressors with low t-statistics.
  - Adjusted R2 used as additional model selection criterion.
  - Changes in macroeconomic variables over a one-year horizon following program approval used as testing horizon.
- Estimation strategy and robustness:
  - OLS used for initial model selection due to lower sensitivity to multicollinearity, errors in variables, or misspecification, particularly in small samples.
  - Endogeneity addressed with 2SLS and 3SLS; probit and tobit used when left-hand-side variables are program interruption dummy and disbursement share, respectively.
  - Instruments considered: country’s IMF quota; cumulative months spent in IMF-supported programs since 1980; amount approved relative to IMF quota; dollar cost of the program starting six months before program approval.
  - 3SLS uses 2SLS estimates to obtain contemporaneous variance-covariance matrix of errors, then applies generalized least squares to a transformed single-equation representation; 3SLS estimates are consistent and asymptotically more efficient than 2SLS.
  - Overidentifying restrictions test could not reject the null for either implementation measure.

- Role of institutions and politics in specification:
  - Domestic institutions and politics treated as key determinants; treated as exogenous and mean-reverting.
  - To avoid collinearity, not all ICRG indices included simultaneously.
  - Institutional influences highlighted:
    - Inflation influenced by law and order, quality of bureaucracy, investment climate; political variation: ethnic tension and internal conflict.
    - Growth affected by investment profile, government stability, initial socioeconomic conditions.
    - Fiscal balance influenced by corruption, democratic accountability, ethnic tension, external conflict, military involvement in politics.
    - Current account influenced by corruption, ethnic tension, external conflict, government stability, investment climate, military participation.

### C. Main Findings — What Determines Program Outcomes?
- First-stage regressions (implementation measures on initial conditions, ICRG indicators, and instruments) used to create predicted values for second-stage.
- Disbursement share as implementation measure:
  - Initial economic conditions: none significantly different from zero.
  - Institutional/political effects:
    - Reduced ethnic tension and greater government stability before approval improve implementation.
      - "Ceteris paribus, a one-point increase in either rating raises disbursements by about 8 percent."
    - Reductions in internal conflict and improvements in law and order in the year before approval associated with lower disbursements:
      - "A one-point increase in the rating [internal conflict] would lower disbursements by just over 6 percent." (internal conflict coefficient negative, significant at the 1 percent level)
      - Law and order rating coefficient negative and significant at the 10 percent level.
    - Greater initial military involvement associated with lower disbursements (ICRG index coefficient positive and significant at the 10 percent level).
  - Program history and size effects:
    - "Every additional month spent in IMF-supported programs translates into 0.2 percent more funds disbursed."
    - Larger programs (amount approved relative to quota) associated with higher disbursement shares.
  - IMF effort measures (staff hours, dollar cost of staff resources): only marginally important.
  - Country’s IMF quota coefficient negative, suggesting weaker implementation in countries with larger Fund quotas.

- Interruption index as implementation measure:
  - Initial economic conditions mostly insignificant; exception: lagged level of growth rate significant (faster preprogram growth → less likely irreversible interruption).
  - Corruption:
    - "A one-point improvement in the ICRG corruption index, all other determinants held constant, coincides with a 35.8 percent better chance of having no program interruption."
  - Internal conflict:
    - Initial level negative and significant; political violence: "an improvement in this rating by one point is associated with a 21.7 percent higher chance of an irreversible interruption."
  - Relationship-with-Fund variables: expected signs but none significantly different from zero except months in program mode.

- Overall interpretation:
  - Implementation largely determined by domestic political economy and institutions; Fund-controlled variables have only marginal impact on program outcomes.

### C. Main Findings — What Determines Macroeconomic Outcomes?
- General estimation details:
  - Dependent variable: change in macroeconomic variable between T - 1 and end of T, T + 1, or T + 2.
  - Tables 5–12 report OLS, 2SLS, and 3SLS; disbursement share or lack of program interruptions used as implementation measures.
  - Focus on 3SLS with disbursement share unless noted.

- Inflation:
  - High persistence: lagged inflation coefficients highly significant for all horizons.
  - Speed of reversal:
    - "For the average program, about three quarters of any deviation from “normal” inflation after a program is approved is reversed within a year."
    - "Deviations of inflation from its long-run equilibrium are erased almost completely in three years."
  - Program implementation:
    - After correcting for endogeneity, disbursement share coefficients negative and declining in absolute value over three years; only first-year result statistically significant.
    - Lack of program interruptions correlated with greater price stability in year following approval, followed by slightly higher inflation over longer horizons.
  - Institutional effects:
    - Better government bureaucracy at program start and improvements subsequently → lower inflation (importance strongest in year one).
    - Better law and order and improvements → lower inflation, though better law and order before start and improvements during program associated with slightly higher inflation at T + 2.
  - Political factors:
    - Lower inflation associated with increased political violence (as proxied by change in ICRG Internal Conflict rating) in first two years; interpretation: demand-side austerity reducing inflation could trigger unrest.
  - Robustness to outliers:
    - Excluding observations with annual change in inflation greater than 50 percent (sample reduced by approximately 30 percent): ethnic tensions remain important; internal conflicts become insignificant; government stability becomes significant.

- Growth:
  - High serial correlation and mean reversion:
    - "Approximately 90 percent of any deviation in growth rates from the country’s “normal” growth pattern is made up within three years."
    - Largest adjustment (83-84 percent) within one year.
  - Program implementation impact:
    - OLS suggests positive and significant association with disbursement share, but 2SLS and 3SLS not significantly different from zero.
    - Impact fragile to implementation measure: fewer interruptions sometimes associated with higher growth in OLS but coefficients turn negative with IV techniques.
    - Possibility that implementation effects on growth occur beyond three-year horizon.
  - Institutional effects:
    - Investment Profile: "A one-point increase in the ICRG Investment Profile rating is associated with roughly a 1 percent increase in the growth rate"—robust across horizons and estimation techniques.
    - Improvements in government stability positively impact growth, especially in first year.
    - Findings robust to implementation measure and omitting outliers (outliers defined as countries growing or shrinking by more than 10 percent per annum).

- Public finances (fiscal balance as percent of GDP):
  - Persistence and mean-reversion:
    - Improvements persist for two years then reverse.
    - "Approximately 45 percent of any deviation of the fiscal balance from long-run average is offset within a year."
  - Program implementation:
    - OLS: implementation associated with improvements in public finances (first two years).
    - IV (2SLS, 3SLS): results reversed; implementation associated with larger fiscal deficits.
      - IV estimates of coefficient on disbursement share two years after approval are significant.
      - "Fiscal deficits in countries with completed programs are about 3 percent larger than in countries whose programs were interrupted."
    - Interpretation: better implementation brings additional financing, allowing more gradual adjustment and larger fiscal deficits initially.
  - Institutional and political influences:
    - Initial level of corruption: most important institutional factor, but effect anomalous—lower corruption associated with weaker fiscal outcomes over time.
    - Democratic Accountability (improvements) associated with larger deficits (possibly democratic incumbents postponing consolidation).
    - Declines in ethnic tension contemporaneously correlated with improved fiscal balances (return to normalcy → improved revenues, lower military spending).
    - Less military involvement preprogram and declines in external conflict associated with lower fiscal deficits.

- Current account:
  - Shocks longer lived; larger permanent components:
    - "Only about 10 percent of any deviation from a country’s “normal” ratio of the current account to GDP is made up for in one year."
  - Program implementation effects:
    - Better implementation associated with deterioration of current account for about two years, followed by sharp improvement in third year.
    - Quantification: "Disbursement of 100 percent of committed funds is accompanied by an 8 percent deterioration of the current account in the first year (relative to the preprogram year), followed by a 2 percent improvement in the third year."
  - Institutional variables:
    - Investment Profile (initial level and change): both highly significant with negative signs—better attitude toward inward investment associated with larger current account deterioration during period.
  - Political variables:
    - External conflict: change in ICRG External Conflict index negative and highly significant for T + 2: "A one-point increase in this rating is correlated with a 1.2 percent deterioration of the current account over three years."
    - Preprogram government stability and less military involvement in politics associated with better current account performance.
  - Results robust to choice of implementation measure.

- Relative contributions across horizons (from empirical decomposition):
  - Program implementation drives inflation in first and second year; institutional and political factors especially important over two-year horizon.
  - For growth over one- and three-year horizons, both program implementation and institutional/political conditions matter, but institutional/political factors have quantitatively greater impact.
  - Institutional factors crucial for fiscal balance over one- and three-year horizons; program implementation very important over two-year horizon.
  - Program implementation major influence on current account in first year; institutional factors stronger over longer horizons.

### IV. Conclusion and Policy Implications
- Four main findings:
  1. Quality of institutions and domestic political environment matter for macroeconomic outcomes in program countries, especially up to three years. Institutional improvements have both direct effects and indirect effects via better program implementation.
     - Examples: better bureaucracy and law enforcement → lower inflation; declines in internal conflict → associated with higher inflation; improved investment profile and government stability → faster growth; easing external conflict and lower military involvement → stronger fiscal outcomes; reductions in ethnic tension and improved accountability → weaker fiscal outcomes; greater government stability and reductions in military involvement → stronger current account; lower ethnic tension and improved investment profile → deterioration of current account.
  2. Institutional and political environment quantitatively important for program implementation. Higher disbursement rates and fewer interruptions where ethnic tensions low, governments stable and less corrupt, and military less involved. More funds disbursed and fewer interruptions where internal conflict was intense and law enforcement weak preprogram—reflecting IMF’s role in facilitating return to normalcy.
  3. Program implementation varies with duration of financial engagement and quota size: more funds disbursed and fewer interruptions in countries with more months in prior IMF programs; better implementation for larger programs (financing relative to quota).
  4. After accounting for institutional impacts, program implementation independently influences macroeconomic outcomes, especially over horizons up to two years:
     - Better implementation associated with lower inflation.
     - Initially weaker but ultimately stronger external and fiscal outcomes.
     - Statistically insignificant impact on economic growth within three years.

- Policy implications:
  - Research agenda:
    - Investigate whether successful program implementation affects growth beyond three-year horizon; study lags in operation of structural reforms.
    - Identify structural reforms that could yield quicker growth payoffs for program design and implementation.
  - Operational suggestions for the IMF:
    - Monitor quantitative institutional and political indicators (e.g., ICRG ratings of ethnic tension, government accountability, investment climate) during program design and implementation to assess prospects for growth and implementation.
    - Use high-frequency information on determinants of investment profile (viability of contracts, threat of expropriation, ease of profit repatriation, payment delays) as implementation feedback.
    - Incorporate quantitative political and institutional analysis into surveillance and program design to provide early warning of declining implementation prospects; consider institutionalizing measures to secure endorsement from major political forces to improve ownership.
  - Modeling recommendations:
    - Move beyond treating institutions as exogenous and mean-reverting; assess systematically the impact of program implementation on institutional and political dynamics by modelling institutional evolution endogenously.

*Source: IMF working paper content (selected sections from the provided PDF content).*

### Box 1. List of Instrumental Variables

### Box 1. List of Instrumental Variables

### Purpose of instrumental variables
- The outcomes of IMF-supported programs are endogenous. Instrumental variables help us obtain unbiased estimates of the impact of IMF-supported programs on the economic performance of participating countries.
- The instruments must be correlated with program implementation (lack of program interruptions and the share of committed funds disbursed) and not be direct determinants of the economic policy outcomes (inflation, economic growth, fiscal balance, and current account).

### List of instrumental variables used in the analysis
- IMF quota (log).
  - A country’s quota determines the member’s voting power in the IMF.
  - Countries with larger quotas have more bargaining power and systemic importance in the world economy.
  - Greater bargaining power could allow countries to extract more concessions from the IMF, leading to less conditionality and more lenient treatment by the Fund. The coefficient on the IMF quota in the implementation measure regressions would then be positive.
  - On the other hand, the size of the quota also reflects a country’s systemic importance in the world economy and its access to international capital markets. Governments of large countries might be less cooperative with IMF conditionality if the perceived political costs are too high. In that case, the parameter estimate could have a negative sign.
- Number of months spent in IMF-supported programs since 1980.
  - This variable captures the extent of a country’s financial involvement with the IMF.
  - The length of the country’s history under Fund-supported programs could lead, through learning-by-doing, to better program design and higher implementation rates as government officials and IMF staff gain more experience and knowledge of country-specific factors and IMF procedures.
- Amount approved as a fraction of IMF quota.
  - This variable is expected to capture the financial importance of a particular program.
  - Large values would be positively correlated with the severity of crises and the willingness of the authorities to implement Fund-supported reforms.
- IMF effort per program year, including six months prior to program approval.
  - This is a direct measure of the dollar cost of IMF-supported programs.
  - It is computed from Budget Reporting System data on hours spent by staff on program implementation and estimated average staff salaries by grade.
  - More effort invested in program implementation is expected to be positively correlated with program implementation.

### Role in empirical strategy
- These instruments are used to identify the causal impact of program implementation measures—specifically lack of program interruptions and the share of committed funds disbursed—on macroeconomic outcomes.
- The instruments are chosen to be correlated with implementation measures while plausibly exogenous to direct determinants of inflation, economic growth, fiscal balance, and current account outcomes.

*Source: Box 1 text from the provided IMF document.*

### conclusions.

### conclusions.

### General findings on determinants and dynamics
- For all four variables, autoregressive and (in most cases) mean-reversion terms are responsible for a large portion of the variation in the explanatory variable at all horizons.
- Initial real GDP per capita is very important for the determination of variation in macroeconomic variables, but the size of this effect declines with the length of the horizon (the only exception is the change in growth rates).
- The marginal impact of political and institutional factors and of program implementation on macroeconomic outcomes varies over time.
- Ratios of the percentage contribution of each of these factors to the contribution of the autoregressive and mean-reversion terms are plotted for inflation, growth, and the fiscal and current account balance in Figure A.1).

### Inflation
- Program implementation drives inflation dynamics in the first and second year of a program.
- Although it works in the same direction as the variable’s autoregressive and mean-reversion dynamics, the impact of program implementation is roughly 3½ and 2 times greater for horizons T and T + 1, respectively.
- When combined, institutional and political factors push inflation upward and are especially important over the two-year horizon (twice the size of the mean-reversion terms).

### Growth
- Institutional factors and program implementation are both important for growth rates over one- and three-year horizons, but institutional and political conditions have a quantitatively greater impact.
- Program effects are growth-stimulating at all horizons (contributing positively to the mean change in growth).
- Institutional and political factors tend to reduce growth in the short run, but raise it in the long run (with a negative contribution to the mean change in growth rates at T, and a positive contribution at T + 2).

### Fiscal balance
- The influences of institutional factors and program implementation are very similar but work in the opposite direction for the fiscal balance.
- Institutional factors are especially important in determining the evolution of the fiscal balance over one- and three-year horizons.
- IMF program implementation is very important over the two-year horizon.

### Current account
- Program implementation is the major factor influencing the current account in the first year of a program.
- The impact is very large, more than seven times greater than the impacts of the autoregressive and mean-reversion factors.
- It appears that program implementation worsens the current account in the short run but improves it in the long run.
- Although institutional factors have a strong impact on current account dynamics over all horizons, their relative importance increases with the length of the horizon.

*Source : Authors’ calculations.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04184.pdf_