## 1.      Previous studies by the staff and the Independent Evaluation Office (IEO) have

## Source details

**Canonical URL:** [1.      Previous studies by the staff and the Independent Evaluation Office (IEO) have](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2012/_072612a.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2012/_072612a.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2012/_072612a.pdf.json)

---

### Literature review and empirical findings
- Prior evaluations found mixed but generally positive effects of Fund engagement on LICs‘ macroeconomic performance:
  - Bredenkamp and Schadler (1999): SAF and ESAF users during 1986–1995 materially strengthened their economies.
  - IEO (2002): IMF lending appeared to have negative effects on growth for some prolonged GRA users (1975–1999), though not for concessional-window users.
  - Ghosh and others (2005): ESAF/PRGF and GRA users (1995–2003) often saw marked reductions in inflation and higher economic growth.
  - Clements, Gupta, and Nozaki (2011): IMF-supported programs had a positive effect on education and health outlays (1985–99).
  - 2011 Review of Conditionality: programs initiated during 2002–11 generally helped improve macroeconomic and social conditions.
- Academic evidence heterogeneous by methodology and sample (selected results):
  - Przeworski and Vreeland (2000): program participation lowers growth (1970–1990; all IMF lending facilities).
  - Barro and Lee (2005): negative impact on growth (1975–1999; short-term stabilization focus).
  - Dicks-Mireaux, Mecagni, and Schadler (2000): beneficial effects on output growth and debt service ratio but no effect on inflation (LICs, 1986–1991).
  - Atoyan and Conway (2006): little contemporaneous growth improvement; stronger evidence in years following a program; fiscal surplus/GDP and current-account surplus/GDP improved contemporaneously (1993–2002).
  - Biglaiser and DeRouen (2010): IMF borrowers more attractive to U.S. investors; effects vary by loan duration, borrower input, and loan amounts (1980–2003).

### Methodological challenges highlighted
- Sample selection bias:
  - Program observations come from a restricted, nonrandom sample; countries often self-select when facing economic difficulties.
  - Countries with weaker initial conditions and institutional weaknesses are more likely to seek IMF support; failing to correct for selection bias yields flawed conclusions.
  - Studies employ various approaches to construct credible counterfactuals and account for start-of-program conditions and country fixed effects.
- Typical empirical shortcomings:
  - Mixing LICs and middle-income economies obscures distinct nature/objectives of Fund engagement.
  - LIC-specific differentiators:
    - (i) Nature of shocks: LICs more vulnerable to frequent domestic and external shocks.
    - (ii) Access to financing: LICs have less access to domestic/external financing and depend on donor assistance and Fund-supported programs.
    - (iii) Longer-term challenges: LIC programs emphasize medium- or longer-term objectives (poverty reduction, capacity- and institution-building).
  - Repeated/continuous Fund engagement in many LICs means snapshot before/after analyses can miss longer-term progress.
- Instrument heterogeneity:
  - Fund instruments differ (medium-term support, episodic short-term and emergency financing, precautionary financing, pure policy support); distinguishing instrument type is important for impact assessment.

### Empirical approach of this study
- Objective: focus on LICs and distinguish short-term versus longer-term instruments to address methodological issues.
- Sectional objectives:
  - Section II: assess longer-term impact of successive medium-term programs (ECF and predecessors; PSI more recently) on growth and socio-economic outcomes using decade averages.
  - Section III: assess short-term financing (augmentations, short-term and emergency instruments) on short-term macro performance—whether financing alleviates liquidity constraints and cushions shocks.
  - Method: Propensity Score Matching (PSM) to control for selection bias (Annex I).
  - Section IV: general conclusions about economic impact of Fund support through LIC facilities.

### Data, sample, and definitions
- Country sample: up to 75 of the 78 countries eligible for concessional support as of January 1, 2010 (Timor-Leste, Somalia, Tonga excluded for lack of data).
- LIC definition: eligible for subsidized resources as of January 1, 2010.
- Timeframe and averaging:
  - Panel of 75 LICs; decadal averages spanning 1986–2010.
  - Decadal periods with 50 percent overlap: 1986–95; 1991–00; 1996–05; 2001–10.
  - Longer-term Fund engagement dummy = 1 if country had five or more years of Fund-supported programs in the 10-year period, zero otherwise.
- Qualifying programs: ECF and predecessors (PRGF, ESAF, SAF), SBA, ESF-HAC, SCF, PSI.
- Program years purged of prolonged interruptions (delays > six months in completing a review owing to noncompliance; interruptions series from Bal Gunduz (2009), updated for 2008–11).

### PSM selection equation and variables
- PSM selection equation estimates determinants of longer-term IMF engagement.
- Independent variables include:
  - Initial reserve coverage and foreign aid to GDP ratio at start of decade.
  - Geographic and institutional dummies (including CPIA-related measures).
  - Trading partners‘ real GDP growth.
  - IMF quota (quota/GDP).
- Aim: parsimonious set achieving relatively good fit (see Annex I for estimation results).

### Stylized facts and descriptive outcomes
- Broad macro improvements across LICs over last two decades:
  - Increases in: real GDP per capita growth, government balances, reserves, current account balances, FDI, exports, institutional quality, social spending.
  - Reductions in: economic volatility, inflation, external debt, poverty.
- LICs with extensive IMF engagement (longer-term users) had, on average, stronger improvements:
  - Larger increases in real GDP per capita growth, government balance, exports, FDI, social spending.
  - Larger reductions in volatility, inflation, and external debt.
  - Current account for longer-term engagement LICs appears to have deteriorated, but current account adjusted for FDI significantly improved (suggesting high import content of FDI).

### Empirical strategy summary
- Two main empirical threads:
  - PSM two-stage approach for longer-term engagement: estimate propensity score; compare 10-year average outcomes between longer-term users and matched non-users.
  - Panel regressions on 10-year averages controlling for determinants of long-run growth and a longer-term engagement dummy to identify transmission channels (macroeconomic stabilization, institutional development, provision of development financing).

---

### PSM results and key econometric outcomes (selected findings)
- PSM indicates longer-term Fund engagement associated with improved macroeconomic and socio-institutional outcomes:
  - Higher long-term real per capita GDP growth in levels and changes.
  - Poverty gap significantly lower and decreases more for longer-term Fund engagement countries.
  - Growth volatility and external debt are lower and CPIA is higher for longer-term engagement countries (changes not always statistically significant versus controls).
  - Changes in inflation, tax revenue, FDI, and social spending are significantly higher in absolute terms for longer-term Fund users while levels often are not significantly different—consistent with longer-term users closing prior performance gaps.
  - Government balance, reserve coverage and their changes positive but not significantly different from control group.

### Panel regressions: methodology and identification
- Two-way fixed-effects panel growth specification used complementarily to PSM.
- Regressions include growth determinants unlikely to be influenced by IMF programs; subsequent specifications add candidate transmission channels (Y) identified from PSM.
- Endogeneity of longer-term engagement addressed via inverse Mills ratio from first-stage PSM selection equation.
- Transmission channel flagging rule: Y is a likely channel if Y is significant and coefficient on longer-term engagement dummy decreases in size and/or significance when Y is included.

### Key regression findings: longer-term Fund engagement and growth
- Panel regressions corroborate PSM: longer-term Fund engagement supports higher real per capita GDP growth.
- Significant transmission channels across specifications include:
  - Inflation
  - Real per capita GDP growth volatility
  - Debt (external debt)
  - Institutional quality
- When controlling for both longer-term engagement and net IMF disbursements in the decade, only the longer-term engagement dummy is significant — implying policy support matters more than financing size for longer-term growth performance.

### Table 1 (selected coefficients and standard errors)
- Growth: Coefficient 1.83*** (St. Error (0.48))
- Change in growth: Coefficient 3.448*** (St. Error (0.78))
- Growth volatility: Coefficient -0.38*** (St. Error (0.13))
- Change in growth volatility: Coefficient -1.747 (St. Error (1.14))
- Inflation: Coefficient -4.192 (St. Error (14.65))
- Change in inflation: Coefficient -14.072*** (St. Error (4.15))
- Tax revenue: Coefficient 0.462 (St. Error (1.25))
- Change in tax revenue: Coefficient 1.67* (St. Error (0.96))
- Government balance: Coefficient 0.952 (St. Error (0.66))
- Change in government balance: Coefficient 1.551 (St. Error (1.65))
- Reserve coverage: Coefficient 0.236 (St. Error (0.38))
- Change in reserve coverage: Coefficient 0.256 (St. Error (0.41))
- Current account: Coefficient -1.593 (St. Error (1.12))
- Change in current account: Coefficient 0.401 (St. Error (1.38))
- FDI: Coefficient 0.875 (St. Error (0.86))
- Change in FDI: Coefficient 1.45** (St. Error (0.66))
- External debt: Coefficient -25.986** (St. Error (13.27))
- Change in external debt: Coefficient -33.559 (St. Error (17.36))
- CPIA: Coefficient 0.302*** (St. Error (0.12))
- Change in CPIA: Coefficient 0.008 (St. Error (0.19))
- Poverty gap: Coefficient -7.47** (St. Error (3.24))
- Change in poverty gap: Coefficient -4.992* (St. Error (3.06))
- Social spending: Coefficient 0.653 (St. Error (0.59))
- Change in social spending: Coefficient 1.602*** (St. Error (0.63))
- Notes: Bootstrapped standard errors in parentheses. Each coefficient represents a separate estimation. Analysis based on four 10-year period averages between 1986 and 2010 with 50 percent overlap. Longer-term engagement defined as five or more years of a financial arrangement or a PSI in a 10-year period.

### Table 2 (selected determinants of long-term real per capita GDP growth)
- Long-term Fund engagement: 0.808** (0.40); 0.794** (0.40); 0.757** (0.38); 0.550 (0.38); 0.813** (0.40); 0.842** (0.40); 0.566 (0.36)
- Inflation: -0.001** (0.00); -0.002*** (0.00)
- Growth volatility: -0.171*** (0.05); -0.102** (0.05)
- Institutional quality: 1.656*** (0.52); 1.742*** (0.61)
- Debt: -0.006* (0.00); 0.009 (0.01)
- IMF disbursements: -0.585 (0.77); -1.073 (0.74)
- Trade openness: 0.030* (0.02); 0.031* (0.02); 0.03* (0.02); 0.034* (0.02); 0.028 (0.02); -0.002 (0.01); 0.008 (0.01)
- Education: 0.044*** (0.02); 0.042*** (0.02); 0.035** (0.02); 0.039*** (0.02); 0.042*** (0.02); 0.03* (0.02); 0.018 (0.02)
- Inverse Mills ratio: -1.639** (0.72); -1.458** (0.71); -1.486** (0.69); -1.619** (0.72); -1.481** (0.72); -2.164*** (0.68); -1.928*** (0.64)
- Constant terms vary across regressions (examples: -2.97 (2.16); -8.198*** (2.43))
- Observations: 234 (most specifications); 199 (some specifications)
- Notes: Robust standard errors in parentheses. Period dummies included. Analysis based on four 10-year period averages between 1986 and 2010 with 50 percent overlap.

### Robustness checks
- PSM: coefficient significance broadly similar across alternative matching approaches; results similar when yearly rolling decadal averages used to increase observations.
- Panel regressions: including foreign aid did not change results substantially; aid not statistically significant.

---

### Impact of short-term Fund financing in LICs: scope and approach
- Focus: short-term Fund support addressing immediate balance of payments needs via augmentations or short-term/emergency instruments.
- Sample and methods: PSM applied to annual probability of participation in Fund-supported programs; selection model finds participation likelihood increases with deterioration in pre-shock macro conditions and magnitude of adverse external shocks.
- Sample period and arrangements: 1980–2010; included SBAs, SAF/ESAF/PRGF/ECF augmentations, CFFs, ESFs, SCFs, RCFs.

### Empirical results: short-term programs for LICs (PSM outcomes and magnitudes)
- Program countries show significantly better outcomes than matched controls on several dimensions: higher growth, current account balances, reserve coverage; lower inflation and fiscal deficits.
- Impacts larger for countries with high propensity scores (PS).
- Reported impacts (All LICs / PS>0.5 / PS>0.7) — Macroeconomic outcomes:
  - Real GDP growth (%): 0.8 / 1.5** / 1.9***
  - Inflation (%): -16.5*** / -20.4*** / -22.2***
  - Reserve coverage (in months of imports): 0.6*** / 0.8*** / 0.7***
  - Current account balance plus FDI (% of GDP): 1.2 / 2.5** / 2.9**
  - Government balance (% of GDP): 1.2** / 1.7** / 1.9**
  - Change in real health spending per capita (%): 8.4* / 8.6 / 13.4
  - Change in real education spending per capita (%): 0.5 / -2.3 / -4.1
  - Change in REER: -3.4 / -3.4 / -5.3
- Change-in-outcomes (X(t)-X(t-1)):
  - Real GDP Growth (%): 0.3 / 1.2 / 1.9**
  - Inflation (%): -6.2* / -10.0** / -12.5**
  - Reserve coverage (in months of imports): 0.6*** / 0.8*** / 0.7***
  - Current account balance plus FDI (% of GDP): 2.2*** / 3.2*** / 3.6***
  - Government balance (% of GDP): 0.6 / 0.8 / 0.9*
- # of observations reported in source: 1/79 044 8357 (as presented)
- Additional PSM observations:
  - Estimated impacts higher for high propensity score countries.
  - Positive growth impact likely driven by Fund financing and catalytic effects easing short-term adjustment burdens.
  - Program countries often have more depreciated REERs, but differences vs control groups not significant.
  - Change in fiscal balances positive but not significantly different from controls except for very high PS countries.
  - Changes in real health and education spending per capita not statistically different from controls (data limited).

### Sample and coverage for short-term PSM figures
- Sample size: 58 LICs.
- Time coverage: 1980-2010.
- Subsamples: All, PS>0.5, PS>0.7.
- Outcomes shown as first differences; figures illustrate estimated impacts on Real GDP Growth, Inflation, Reserve coverage, Government balance, Current account + FDI, REER change, Real health spending per capita, Real education spending per capita.

---

### Concluding remarks: policy implications and interpretation
- Two distinct beneficial channels of Fund support for LICs:
  - Longer-term policy support (ECF, PSI and predecessors): associated with higher long-term growth rates, less growth volatility, faster poverty reduction, higher tax ratios, higher social spending, improved institutional quality, higher FDI, and lower inflation and external debt; effects appear independent of the amount of Fund financing over the longer term.
  - Short-term liquidity support (augmentations, SCF, RCF, etc.): associated with higher short-term growth, improved current account balances and reserve coverage, and lower inflation and fiscal deficits for countries with immediate BOP needs.
- Implications for Fund instruments:
  - LIC facilities should include both medium-term policy support tools (e.g., ECF, PSI) and rapid financing options (e.g., augmentations, SCF, RCF) to provide appropriate policy support and timely liquidity.
  - Precautionary arrangements (ECF, precautionary SCF) can provide policy support and insurance value even at low access levels in the absence of shocks.
- Historical context:
  - Combination of stronger pre-crisis buffers from longer-term engagement and the Fund’s sharp increase in financial assistance in 2009 helped LICs mount a countercyclical fiscal response in 2009 and contributed to rapid economic recovery.

### ANNEX I — PSM methodology: key points and selection model results
- PSM approach used: two-step (selection model = pooled probit; matching = nearest neighbor with three nearest neighbors).
- Identification and assumptions:
  - Conditional independence and common support required.
  - ATT estimated as mean difference between matched treatment and control outcomes.
- Selection model for longer-term engagement (pooled probit; observations: 236):
  - Initial reserves: -0.166*** (0.04)
  - Initial aid/GDP: 0.025*** (0.01)
  - Trading partner growth: -0.129* (0.07)
  - IMF quota/GDP: -0.070** (0.03)
  - Francophone: 0.436** (0.21)
  - Landlocked: 0.787*** (0.20)
  - PRGF reform dummy: -0.771*** (0.21)
  - Constant: 0.571* (0.34)
- Interpretation: higher initial reserves lower propensity of longer-term engagement; higher initial aid/GDP and being francophone or landlocked increase propensity; larger IMF quota implies lower propensity.
- Selection model for short-term engagement (correlated random effects probit; observations: 532; # countries: 55; Pseudo R2 0.58; sample probability 0.44):
  - Current account balance to GDP (t-1): -0.076*** (-4.61)
  - Reserve coverage (CFA zone) (t-1): -0.478*** (-6.08)
  - Reserve coverage (non-CFA) (t-1): -0.769*** (-8.71)
  - Macroeconomic stability indicator (t-1): 0.068*** (2.89)
  - Real GDP growth (t-1): -0.113*** (-4.24)
  - Change in terms of trade (t-1): -0.022*** (-2.8)
  - Change in real oil prices (previous two years): 0.009*** (2.85)
  - Real world trade, cyclical component: -0.099** (-2.53)
  - Paris Club dummy: 0.774*** (3.24)
- Interpretation: worse reserve coverage, worse current account, lower growth, adverse terms-of-trade and global shocks increase probability of requesting Fund financing.

*Italic: Source document _072612a - Panel regressions and PSM results (excerpts drawn from the provided PDF).*

### 1.      Previous studies by the staff and the Independent Evaluation Office (IEO) have

### _072612a - 1.      Previous studies by the staff and the Independent Evaluation Office (IEO) have

### Literature review and empirical findings
- Previous staff and IEO studies identified some positive effects of Fund engagement on LICs’ macroeconomic performance, but not in all areas:
  - Bredenkamp and Schadler (1999): reviewed countries using the Fund‘s concessional lending window under three-year SAF and ESAF arrangements during 1986–1995 and concluded most countries pursuing reforms with SAF and ESAF strengthened their economies materially.
  - IEO report (2002): mixed sample using GRA and concessional Fund resources over 1975–1999; econometric methods found IMF lending appeared to have negative effects on growth for some prolonged users of Fund resources, though not for those using the Fund‘s concessional lending window.
  - Ghosh and others (2005): reviewed LICs with ESAF and PRGF and GRA users over 1995–2003; found countries with IMF-supported programs often experienced marked reductions in inflation and higher economic growth.
  - Review of Ex Post Assessments and Issues Relating to the Policy on Longer-Term Engagement (inputs from EPA teams, mission chiefs, country authorities): macroeconomic problems reduced in many countries with longer-term program engagement, while structural problems often persisted.
  - Clements, Gupta, and Nozaki (2011): econometric evidence based on LICs and GRA users for 1985–99 found IMF-supported programs had a positive effect on education and health outlays among LICs.
  - 2011 Review of Conditionality—Outcomes of Fund-Supported Programs: examined programs initiated during 2002–11 through multiple approaches and concluded most Fund-supported programs appear to have helped member countries improve macroeconomic and social conditions.
- Academic literature shows mixed results depending on methodology and sample:
  - Przeworski and Vreeland (2000): program participation lowers growth (1970–1990; all IMF lending facilities).
  - Barro and Lee (2005): negative impact on growth (1975–1999; short-term stabilization focus).
  - Dicks-Mireaux, Mecagni, and Schadler (2000): beneficial effects on output growth and debt service ratio but no effect on inflation (LICs, 1986–1991).
  - Atoyan and Conway (2006): little contemporaneous improvement in real economic growth but stronger evidence of improvement in years following a program; fiscal surplus/GDP and current-account surplus/GDP improved contemporaneously (developing and transition countries, 1993–2002; GRA and LIC facilities).
  - Biglaiser and DeRouen (2010): IMF borrowers tend to be more attractive to U.S. investors, but effects vary by loan duration, borrower input in policy decisions, and loan amounts (mixed LICs and emerging economies, 1980–2003).

### Methodological challenges highlighted
- Sample selection bias:
  - Economic performance under Fund-supported programs is observed only for a restricted, nonrandom sample of countries with programs; countries may self-select (e.g., seek IMF because already facing economic difficulties).
  - Countries with weaker initial conditions or institutional weaknesses are more likely to seek IMF support; failing to correct for selection bias leads to flawed conclusions.
  - Studies use various approaches to construct credible counterfactuals; the literature commonly aims to account for conditions at the start of a program and unique country characteristics.
- Typical empirical shortcomings:
  - Mixing LICs and middle-income economies obscures distinct nature and objectives of Fund engagement across country groups.
  - LIC-specific differentiators:
    - (i) Nature of shocks: LICs more vulnerable to frequent domestic and external shocks reflecting lack of development and diversification.
    - (ii) Access to financing: LICs have less access to domestic/external financing, depend on donor assistance and Fund-supported programs to catalyze assistance.
    - (iii) Longer-term challenges: LIC programs emphasize medium- or longer-term objectives important for poverty reduction and growth; emphasize capacity- and institution-building beyond short-term policies.
  - Many LICs experience continuous or repeated Fund engagement; snapshot analyses before/after individual programs can miss progress toward longer-term objectives.
- Instrument heterogeneity not always distinguished:
  - Fund instruments vary (medium-term support, episodic short term and emergency financing, precautionary financing, pure policy support).
  - Medium-term instruments emphasize entrenched imbalances and institutional weaknesses; short-term instruments focus on financing and adjustment to shocks.
  - Differentiation by instrument type has important implications for impact assessment.

### Empirical approach of this study
- Aim: address methodological considerations by focusing on LICs and differentiating between short-term and longer-term instruments.
- Sectional objectives summarized:
  - Section II: assess longer-term impact of Fund engagement—primarily through successive medium-term programs under the ECF and its predecessors (and more recently the PSI)—on longer-term economic growth and other indicators and socio-economic outcomes; assess longer-term impact by looking at decade averages rather than contemporaneous indicators.
  - Section III: assess impact of short-term financing (augmentations of ECF arrangements and short-term and emergency financing instruments) on short-term macroeconomic performance, focusing on whether short-term Fund financing alleviates liquidity constraints and cushions shocks.
  - For both analyses use the Propensity Score Matching (PSM) approach to control for selection bias (Annex I for details).
  - Section IV: suggest general conclusions about economic impact of Fund support through LIC facilities.

### Data, sample, and definitions
- Country sample:
  - Up to 75 of the 78 countries eligible for concessional support as of January 1, 2010; Timor-Leste, Somalia, and Tonga excluded for lack of data.
  - LICs defined as those eligible for subsidized resources as of January 1, 2010.
- Timeframe and averaging:
  - Panel dataset of 75 LICs; decadal averages spanning 1986–2010.
  - Decadal periods share a 50 percent overlap: 1986–95; 1991–00; 1996–05; 2001–10.
  - For any 10-year period, longer-term Fund engagement dummy = 1 if country had five or more years of Fund-supported programs in the 10-year period and zero otherwise.
- Qualifying programs: all Fund financial arrangements available to LICs, primarily ECF and predecessors (PRGF, ESAF, SAF) and also SBA, ESF-HAC, SCF, and the PSI.
- Program years purged of episodes with prolonged program interruptions (examined via delays > six months in completing a review owing to noncompliance with macroeconomic performance criteria; interruptions series from Bal Gunduz (2009), updated for 2008–11).

### PSM selection equation and variables
- PSM selection equation estimates determinants of longer-term IMF engagement.
- Independent variables broadly align with literature on demand and supply factors for IMF support:
  - Initial macroeconomic buffers proxied by reserve coverage and foreign aid to GDP ratio at start of each decade.
  - Structural characteristics proxied by geographic and institutional dummy variables (including CPIA-related institutional measures).
  - Trading partners‘ real GDP growth captures external demand conditions (exogenous to LICs).
  - Countries‘ access to IMF resources proxied by IMF quota.
- Aim: parsimonious set of variables achieving relatively good fit based on historical data series.
- See Annex I for additional discussion and estimation results.

### Stylized facts and descriptive outcomes
- Macro improvements over last two decades for most LICs (Figure 1 referenced):
  - Increases in: real GDP per capita growth, government balances, reserves, current account balances, FDI, exports, institutional quality, social spending.
  - Reductions in: economic volatility, inflation, external debt, poverty.
  - Finding holds across country sizes, geographical groupings, institutional capacity (World Bank CPIA), and per-capita income (Figure 2 referenced).
- LICs with extensive IMF program engagement have, on average, comparatively stronger improvements:
  - Extensive users faced weaker initial conditions in the 1980s but experienced larger increases in real GDP per capita growth, government balance, exports, FDI, and social spending than countries without extensive engagement.
  - More marked reduction in economic volatility, inflation, and external debt.
  - Strong improvement largely eliminated the performance gap relative to other LICs around ESAF creation in 1987.
  - Note: current account for LICs with longer-term engagement seems to have deteriorated over last two decades, but current account adjusted for FDI significantly improved (suggesting likely high import content of FDI).

### Econometric analysis objectives and methods
- Questions addressed:
  - a. How does longer-term Fund engagement affect macroeconomic performance, including growth, and institutional variables? Method: PSM two-stage process—(i) estimate propensity score of becoming longer-term user; (ii) compare average economic performance over 10-year period between longer-term users and others with similar propensity scores.
  - b. What is the longer-term impact of IMF engagement on long-term economic growth and associated transmission channels? Method: panel regressions on 10-year averages controlling for determinants of long-run growth and a dummy for longer-term Fund engagement to identify channels: macroeconomic stabilization, institutional development, provision of development financing.
- Data construction choices:
  - Decadal averages used (periods overlapping as above) to capture longer-term engagement and reduce bias from arbitrary period selection.
  - Longer-term engagement defined as five or more years of financial arrangement or PSI in a 10-year period.
  - Qualifying arrangements listed above; program years purged for prolonged interruptions.

### PSM results and key econometric outcomes
- PSM suggests longer-term Fund engagement associated with improved macroeconomic and socio-institutional outcomes (Table 1 referenced). Key findings:
  - Longer-term Fund engagement leads to significantly higher long-term real per capita GDP growth both in levels and changes.
  - Poverty gap is significantly lower and decreases more for countries with longer-term Fund engagement.
  - Growth volatility and external debt are lower and CPIA is higher for countries with longer-term engagement, although changes in these variables are not significantly different from control group.
  - Changes in inflation, tax revenue, FDI, and social spending are significantly higher in absolute terms for longer-term Fund users while the corresponding levels are not significantly different from control group—corroborates that strong improvement of longer-term users largely eliminated performance gaps over 1986–2010.
  - Government balance and its changes are positive but not significantly different from control group; possible explanation: LICs are financing-constrained and Fund-supported programs provide direct financing and catalyze donor support allowing them to address development needs.
  - Reserve coverage and its changes are positive but not significantly different from control group.

*Italic: Source document _072612a - 1.      Previous studies by the staff and the Independent Evaluation Office (IEO) have*

### 13.      Panel regressions are used as a complementary approach to estimate the impact

### _072612a - 13.      Panel regressions are used as a complementary approach to estimate the impact

### Methodology: panel regressions and identification
- Two-way fixed-effects panel growth specification used as a complementary approach to Propensity Score Matching (PSM).
- Initial regressions include growth determinants unlikely to be directly influenced by IMF-supported programs; subsequent regressions augment the specification with variables identified as likely transmission channels from the PSM analysis.
- A variable is considered a likely transmission channel if it is significant and the coefficient associated with the longer-term Fund engagement dummy decreases in size and/or significance relative to the benchmark model.
- Endogeneity of longer-term Fund engagement is addressed through the inverse Mills ratio estimated in the first-stage PSM selection equation model.

### Key regression findings: longer-term Fund engagement and growth
- Panel growth regressions corroborate the PSM finding that longer-term Fund engagement appears to support higher real per capita GDP growth.
- Transmission channels identified as significant across specifications include:
  - Inflation
  - Real per capita GDP growth volatility
  - Debt (external debt)
  - Institutional quality
- When controlling for both longer-term Fund engagement and the size of net IMF disbursements in the decade, only the longer-term Fund engagement dummy is significant — implying policy support matters more than financing size for longer-term growth performance.

### Table 1: Impact of Longer-term Fund Engagement on Economic and Institutional Performance (selected coefficients and standard errors)
- Growth: Coefficient 1.83*** (St. Error (0.48))
- Change in growth: Coefficient 3.448*** (St. Error (0.78))
- Growth volatility: Coefficient -0.38*** (St. Error (0.13))
- Change in growth volatility: Coefficient -1.747 (St. Error (1.14))
- Inflation: Coefficient -4.192 (St. Error (14.65))
- Change in inflation: Coefficient -14.072*** (St. Error (4.15))
- Tax revenue: Coefficient 0.462 (St. Error (1.25))
- Change in tax revenue: Coefficient 1.67* (St. Error (0.96))
- Government balance: Coefficient 0.952 (St. Error (0.66))
- Change in government balance: Coefficient 1.551 (St. Error (1.65))
- Reserve coverage: Coefficient 0.236 (St. Error (0.38))
- Change in reserve coverage: Coefficient 0.256 (St. Error (0.41))
- Current account: Coefficient -1.593 (St. Error (1.12))
- Change in current account: Coefficient 0.401 (St. Error (1.38))
- FDI: Coefficient 0.875 (St. Error (0.86))
- Change in FDI: Coefficient 1.45** (St. Error (0.66))
- External debt: Coefficient -25.986** (St. Error (13.27))
- Change in external debt: Coefficient -33.559 (St. Error (17.36))
- CPIA: Coefficient 0.302*** (St. Error (0.12))
- Change in CPIA: Coefficient 0.008 (St. Error (0.19))
- Poverty gap: Coefficient -7.47** (St. Error (3.24))
- Change in poverty gap: Coefficient -4.992* (St. Error (3.06))
- Social spending: Coefficient 0.653 (St. Error (0.59))
- Change in social spending: Coefficient 1.602*** (St. Error (0.63))
- Notes: Bootstrapped standard errors in parentheses. Each coefficient represents a separate estimation. Analysis based on four 10-year period averages between 1986 and 2010 with 50 percent overlap. Definition: a country is considered to have longer-term(LT) engagement in a given decade if in five or more years it had a financial arrangement or a PSI in place, for at least six months in each of these years. Changes refer to first decadal differences.

### Table 2: Determinants of Long-term Real per Capita GDP Growth (selected coefficients and standard errors)
- Long-term Fund engagement: 0.808** (0.40); 0.794** (0.40); 0.757** (0.38); 0.550 (0.38); 0.813** (0.40); 0.842** (0.40); 0.566 (0.36)
- Inflation: -0.001** (0.00); -0.002*** (0.00)
- Growth volatility: -0.171*** (0.05); -0.102** (0.05)
- Institutional quality: 1.656*** (0.52); 1.742*** (0.61)
- Debt: -0.006* (0.00); 0.009 (0.01)
- IMF disbursements: -0.585 (0.77); -1.073 (0.74)
- Trade openness: coefficients include 0.030* (0.02); 0.031* (0.02); 0.03* (0.02); 0.034* (0.02); 0.028 (0.02); -0.002 (0.01); 0.008 (0.01)
- Education: 0.044*** (0.02); 0.042*** (0.02); 0.035** (0.02); 0.039*** (0.02); 0.042*** (0.02); 0.03* (0.02); 0.018 (0.02)
- Inverse Mills ratio: -1.639** (0.72); -1.458** (0.71); -1.486** (0.69); -1.619** (0.72); -1.481** (0.72); -2.164*** (0.68); -1.928*** (0.64)
- Constant terms include -2.97 (2.16); -2.968 (2.14); -1.768 (2.12); -8.198*** (2.43); -2.529 (2.19); 1.482 (1.99); -4.089 (3.07)
- Observations: 234; 234; 234; 234; 234; 199; 199
- Notes: Robust standard errors in parentheses. Period dummies included (not shown). Analysis based on four 10-year period averages between 1986 and 2010 with 50 percent overlap. Longer-term engagement defined as in Table 1.

### Robustness checks
- PSM: coefficient significance broadly similar across alternative matching approaches; results similar when yearly rolling decadal averages used to increase observations.
- Panel regressions: including foreign aid did not change results substantially; aid was not statistically significant.

### Impact of short-term Fund financing in LICs: scope and approach
- Focus: short-term Fund financial support addressing immediate balance of payments needs arising from policy slippages and/or external shocks, via augmentations of access under existing arrangements or short-term/emergency financing instruments.
- Sample and methods: PSM methodology applied to annual probability of participation in Fund-supported programs; selection model finds participation likelihood increases with deterioration in pre-shock macroeconomic conditions and magnitude of adverse external shocks.
- Sample period and arrangements: sample period covers 1980–2010. Included arrangements: SBAs, SAF/ESAF/PRGF/ECF augmentations, Compensatory Financing Facilities (CFFs), ESFs, SCFs, and Rapid Credit Facilities (RCFs). (More details provided in Annex I of source.)

### Empirical results: short-term programs for LICs (PSM outcomes and magnitudes)
- Program countries exhibit significantly better outcomes than matched control groups on several dimensions:
  - Higher growth, current account balances, and reserve coverage.
  - Lower inflation and fiscal deficits.
- Estimated impacts larger for countries with high propensity scores (indicating severe pre-existing imbalances and/or large shocks).
- Specific reported figures (All LICs / PS>0.5 / PS>0.7) — Macroeconomic outcomes:
  - Real GDP growth (%): 0.8 / 1.5** / 1.9***
  - Inflation (%): -16.5*** / -20.4*** / -22.2***
  - Reserve coverage (in months of imports): 0.6*** / 0.8*** / 0.7***
  - Current account balance plus FDI (% of GDP): 1.2 / 2.5** / 2.9**
  - Government balance (% of GDP): 1.2** / 1.7** / 1.9**
  - Change in real health spending per capita (%): 8.4* / 8.6 / 13.4
  - Change in real education spending per capita (%): 0.5 / -2.3 / -4.1
  - Change in REER: -3.4 / -3.4 / -5.3
- Change in macroeconomic outcomes (X(t)-X(t-1)):
  - Real GDP Growth (%): 0.3 / 1.2 / 1.9**
  - Inflation (%): -6.2* / -10.0** / -12.5**
  - Reserve coverage (in months of imports): 0.6*** / 0.8*** / 0.7***
  - Current account balance plus FDI (% of GDP): 2.2*** / 3.2*** / 3.6***
  - Government balance (% of GDP): 0.6 / 0.8 / 0.9*
- # of observations: 1/79 044 8357 (as presented in source)
- Additional PSM findings:
  - Estimated impacts are higher for countries with high propensity scores.
  - Positive growth impact likely explained by Fund financing and potential catalytic effects easing short-term adjustment burdens.
  - Program countries often have more depreciated real exchange rates, but differences with control groups are not significant.
  - Change in fiscal balances is positive but not significantly different from control groups except for very high propensity score countries.
  - Changes in real health and education spending per capita are not statistically different from control groups (data limited).

### Concluding remarks: policy implications and interpretation
- Two distinct channels through which Fund support benefits LICs:
  - Longer-term policy support: associated with higher long-term growth rates, less growth volatility, more rapid poverty reduction, higher tax ratios, higher levels of social spending, higher institutional quality, higher FDI, and lower inflation and external debt; effect appears independent of amount of Fund financing over the longer term.
  - Short-term liquidity support: associated with higher short-term growth, improved current account balances and reserve coverage, and lower inflation and fiscal deficits for countries facing immediate balance of payments needs.
- Implications for Fund instruments:
  - Facilities for LICs should include both medium-term policy support tools (e.g., ECF, PSI) and rapid financing options (e.g., augmentations, SCF, RCF) to provide appropriate policy support and timely liquidity.
  - Precautionary arrangements (ECF, precautionary SCF) can provide policy support and insurance value even at low access levels in the absence of shocks.
- Historical context:
  - Combination of stronger pre-crisis buffers from longer-term engagement and the Fund’s sharp increase in financial assistance in 2009 helped LICs mount a countercyclical fiscal policy response in 2009 and contributed to rapid economic recovery.

*Italic source attribution: IMF staff analysis from content unit _072612a - Panel regressions and PSM results (excerpts drawn from the provided PDF).*

### ANNEX I. PROPENSITY SCORE MATCHING (PSM) METHODOLOGY

### ANNEX I. PROPENSITY SCORE MATCHING (PSM) METHODOLOGY

### A. Addressing selection bias — alternative approaches
- Overview of alternative methods to construct a counterfactual:
  - Before-after approach: attributes changes to Fund-supported programs; vulnerable to biases from structural changes or unrelated shocks (Ghosh and others (2005)).
  - Instrumental variables: requires instruments correlated with treatment selection but not with outcomes; identification of truly exogenous instruments is a major challenge (Barro and Lee (2005)).
  - Generalized Evaluation Estimator (GEE): uses policy reaction functions for non-program countries to approximate the counterfactual (Dicks-Mireaux, Mecagni, and Schadler (2000)); identification of exogenous instruments remains challenging.
  - Heckman selection correction: first-stage probit to predict program engagement, second-stage includes inverse Mills‘ ratio (Przeworski and Vreeland (2000)).
  - Propensity Score Matching (PSM): chosen in this supplement (discussion follows).

### B. PSM methodology (approach used)
- Two-step statistical comparison:
  - Step 1 (selection model): estimate the probability of participating in Fund-supported programs conditional on observable economic conditions and country characteristics.
  - Step 2 (matching): use estimated propensity scores to match program countries to non-program countries and construct a statistical control group.
- Matching approach used:
  - Nearest neighbor matching with three nearest neighbors: constructs control group by choosing the three non-program countries with probability of requesting a program as close as possible to that of the program country.
- Identification and assumptions:
  - Conditional independence (confoundedness): program participation must be based entirely on observed pre-shock characteristics of LICs; violation occurs if unobserved characteristics determine participation.
  - Common support: requires treatment observations to have comparable control observations "nearby" in the propensity score distribution (p(Xi) < 1).
- Estimation target:
  - Average treatment effect on the treated (ATT) computed as the mean difference in macroeconomic outcomes between matched treatment and control groups.
- Practical notes:
  - PSM useful when only observed pre-treatment characteristics affect program participation; rich pre-program data and a well-specified selection model are key.
  - Nearest neighbor matching sorts records by estimated propensity score and searches forward and backward for closest control units; study uses three nearest neighbors.

### C. Specification of the PSM selection model
- Motivation and sample construction:
  - Focus on low-income countries (LICs); distinguishes longer-term engagement from short-term financing to create more homogeneous samples and improve identification.
- Selection model for longer-term Fund engagement:
  - Estimation: pooled probit regression.
  - Dependent variable: dummy = 1 if a country has had five or more years of Fund-supported programs in a 10-year period, zero otherwise.
  - Qualifying programs: ECF and predecessors (PRGF, ESAF, SAF), SBA, ESF-HAC, SCF, and the PSI. Program years purged of episodes with prolonged interruptions (interruption = delay > six months in completing a review due to noncompliance).
  - Decadal averages used with 50 percent overlap; decadal periods: 1986–95; 1991–00; 1996–05; 2001–10.
  - Independent variables: initial reserve coverage, foreign aid to GDP ratio, geographic and institutional dummies (e.g., francophone, landlocked), trading partners‘ real GDP growth, IMF quota (quota/GDP).
- Empirical findings (Annex I Table 1):
  - Coefficients and significance (standard errors in parentheses):
    - Initial reserves -0.166*** (0.04)
    - Initial aid/GDP 0.025*** (0.01)
    - Trading partner growth -0.129* (0.07)
    - IMF quota/GDP -0.070** (0.03)
    - Francophone 0.436** (0.21)
    - Landlocked 0.787*** (0.20)
    - PRGF reform dummy -0.771*** (0.21)
    - Constant 0.571* (0.34)
  - Observations: 236
  - Significance notation: * 10 percent significance; **5 percent significance; ***1 percent significance
  - Notes: Robust standard errors in parentheses. Definition: a country is considered to have longer-term (LT) engagement in a given decade if in five or more years it had a financial arrangement or a PSI in place, for at least six months in each of these years.
  - Interpreted outcomes from the model:
    - Higher initial reserves associated with lower propensity of longer-term Fund engagement.
    - Higher initial foreign aid/GDP associated with higher propensity of longer-term engagement.
    - Lower trading partner growth associated with higher propensity of longer-term engagement.
    - Francophone and landlocked countries have higher propensity for longer-term engagement.
    - Larger IMF quota implies a lower probability of longer-term engagement.
    - PRGF reform dummy associated with lower propensity of longer-term engagement.

- Selection model for short-term Fund engagement:
  - Basis: draws on Bal Gündüz (2009); dependent variable is a panel dummy = 1 if a new Fund arrangement is approved (arrangements addressing immediate balance of payments needs), zero otherwise.
  - Included arrangements: SBA, SAF/ESAF/PRGF/ECF augmentations, ESF, SCF, RCF, CFF (with refinements: exclude precautionary SBA/SCF and SBA/PRGF/ECF augmentations addressing natural disasters; include some SAF/ESAF/PRGF/ECF cases addressing immediate balance of payments needs).
  - Normal episodes: initial year of two successive years with no Fund financing when member is eligible to access Fund resources; additional exclusions applied for supply constraints, arrears, ENDA, SMP, EPCA, program interruptions, break-up of negotiations, and episodes with arrears to other creditors without adjustment programs/rescheduling.
  - Estimation: binary response model for panel data (panel probit). Different estimators considered for panel heterogeneity; correlated random effects probit model preferred based on econometric tests for significance of individual specific effect and sample average for covariates.
  - Key empirical determinants of increased probability of Fund financing (Annex I Table 2 coefficients and statistics listed below):
    - Current account balance to GDP (t-1): -0.076*** (-4.61)
    - Reserve coverage in months of imports (CFA Franc zone) (t-1): -0.478*** (-6.08)
    - Reserve coverage in months of imports (non-CFA Franc zone) (t-1): -0.769*** (-8.71)
    - Macroeconomic stability indicator (t-1): 0.068*** (2.89)
    - Real GDP growth (t-1): -0.113*** (-4.24)
    - Change in terms of trade (t-1): -0.022*** (-2.8)
    - Change in real oil prices in previous two years: 0.009*** (2.85)
    - Real world trade, cyclical component: -0.099** (-2.53)
    - Change in real non-oil commodity prices: -0.020 (-1.58)
    - Real growth of goods exports (t-1): -0.009* (-1.79)
    - Paris Club dummy: 0.774*** (3.24)
    - Constant: 0.551 (1.23)
  - Country specific averages (reported):
    - Total debt service to exports: 0.044*** (2.63)
    - FDI to GDP: -0.105* (-1.76)
  - Model fit and sample statistics:
    - Pseudo R2 0.58
    - # of observations 532
    - # of countries 55
    - Sample probability 0.44
  - Interpretation and conclusions:
    - Reserve coverage, current account balance to GDP, real GDP growth, macroeconomic stability indicator, and terms of trade shocks are significantly associated with increased probability of requesting Fund financing.
    - Adverse global shocks (change in real oil and non-oil commodity prices, cyclical component of world trade) increase participation in Fund arrangements.
    - Demand for Fund resources by LICs is likely to be cyclical in response to global conditions, with intensity depending on magnitude and persistence of adverse external shocks.
- Treatment variable and sample refinements for short-term impact analysis:
  - Treatment variable: panel dummy = 1 for approval of Fund-supported programs addressing immediate balance of payments needs, zero for non-program episodes; refinements to program and non-program episodes mirror selection equation refinements.
  - Within program countries, a higher propensity score identifies programs addressing a clear financing need.
  - Severe state failure events excluded from both program and non-program sets (identified from Political Instability Task Force dataset); program interruption years excluded to account for implementation.

*Source: ANNEX I. PROPENSITY SCORE MATCHING (PSM) METHODOLOGY, _072612a - ANNEX I. PROPENSITY SCORE MATCHING (PSM) METHODOLOGY*

### Annex I Table 2: Demand for Fund Financing in Response to Policy

### Annex I Table 2: Demand for Fund Financing in Response to Policy and/or External Shocks

### Demand for Fund financing: estimation and notes
- Source: Staff calculations (Bal Gündüz, 2009).
- Method: Demand for Fund financing in response to policy and/or exogenous shocks excluding natural disasters is estimated by a correlated random effects probit model.
- Statistical significance notation: Significant at 10 percent:*; 5 percent:**; and 1 percent:***, t-statistics in parantheses.
- Country-specific averages: calculated as the sample average of variables for each country.
- LR tests referenced: (Prob) 0 :test LR 2  and (Prob) 0 :test LR 2 2  k .

### Key methodological context (from Annex II)
- Outcome of interest: real GDP per capita growth rate.
- Estimation approach: two-way fixed-effects models for panel data.
- Selection correction: all regression specifications control for the inverse Mills ratio to address selection bias.
- Focus: effect of longer-term Fund engagement on the long-term average real GDP per capita growth rate.
- Model structure: Z is the matrix of control variables chosen not to be related to Fund engagement; inclusion of potential transmission channel variables Y (those significantly affected by the IMF program dummy in PSM estimation) is used to assess channel strength.
- Interpretation rule: If inclusion of Y lowers (in absolute terms) the magnitude and the significance of the coefficient associated with the IMF dummy, Y is identified as a transmission channel of Fund-supported programs.

### Severe state failure indicator
- Variable used: SFTPMMAX presents the maximum magnitude of all events in a year; values exceeding 3.9 are taken as a severe state failure event.

*Source: _072612a - Annex I Table 2: Demand for Fund Financing in Response to Policy and/or External Shocks (Staff calculations, Bal Gündüz, 2009)._

### Macroeconomic Conditions in LICs across Decades

### Sample and definitions
- Sample size: 75 LICs.
- Decadal averages shown: unweighted average (except inflation which shows the median) over each decade.
- Decades covered in figures: 1981-90, 1991-00, 2001-10; overlapping decadal periods for some analyses: 1986-95; 1991-00; 1996-05; 2001-10.
- Definition of longer-term engagement: 10 or more years of having an IMF financial arrangement or PSI in place during 1991-2010, for at least six months in each of these years.

### Macroeconomic indicators tracked (figures and charts)
- Real GDP per capita growth (Annual averages)
- Inflation (Median of averages)
- Government balance (In percent of GDP)
- Current account + FDI (In percent of GDP)
- Reserves (In months of imports)
- Real GDP per capita volatility (Averages of standard deviations)
- Tax revenue (In percent of GDP)
- Capital spending (In percent of GDP)
- Current account (In percent of GDP)
- External debt (In percent of GDP)
- Aid (In percent of GDP)
- Social spending (In percent of GDP)
- CPIA (example levels shown: 2.7, 2.8, 2.9, 3.0, 3.1, 3.2, 3.3, 3.4, 3.5, 3.6, 3.7 for 1990 2000 2009)
- FDI (In percent of GDP)
- Exports (In percent of GDP)
- Poverty (index plotted across decades)
- Education spending (In percent of GDP)
- Health spending (In percent of GDP)

### Country group comparisons shown in figures
- Groupings illustrated: Landlocked, Coastal, Fragile, Non-Fragile, Per capita GDP < 1175, Per capita GDP > 1175, Small economies, Non-small economies.
- Charts compare "LICs with more than 10 program years" versus "All LICs" across decades for each indicator.

*Source: _072612a - Annex I Table 2: Demand for Fund Financing in Response to Policy and/or External Shocks (Figures for macroeconomic conditions; sample of 75 LICs)._

### Changes in Macroeconomic Performance of LICs (Decadal changes and distributions)

### Sample and definition for decadal-change distributions
- Sample composition: 75 LICs.
- Overlapping decadal period averages: 1986-95; 1991-00; 1996-05; 2001-10.
- Definition of longer-term (LT) engagement in a given decade: if in five or more years it had a financial arrangement or a PSI in place, for at least six months in each of these years.
- Chart presentation: distribution of decadal changes across countries by quartiles (Median, 25th percentile, 75th percentile).

### Indicators for decadal change distributions (Figure 3 and continuations)
- Change in average decadal real GDP per capita growth (percent)
- Change in average decadal real GDP per capita volatility (percent)
- Change in average decadal inflation (percent)
- Change in average decadal FDI/GDP
- Change in average decadal government balance/GDP
- Change in average decadal (current account + FDI)/GDP
- Change in average decadal exports/GDP
- Change in external debt/GDP (noted ranges in figure axis: -80 to 20)
- Change in CPIA (axis range shown: -0.6 to 0.6)
- Change in average decadal social spending/GDP (axis range shown: -1.5 to 2.0)
- Change in average decadal poverty gap (axis range shown: -18.0 to 4.0)
- Change in average decadal aid/GDP (axis range shown: -6 to 6)
- Change in average decadal education spending/GDP (axis range shown: -1.0 to 1.6)
- Change in average decadal tax revenue/GDP (axis range shown: -1.5 to 3.5)
- Change in average decadal capital spending/GDP (axis range shown: -3 to 4)
- Change in average decadal current account/GDP (axis range shown: -5 to 5)
- Change in average decadal reserve coverage (In months of imports) (axis range shown: -0.5 to 3.0)
- Change in average decadal health spending/GDP (axis range shown: -0.8 to 1.0)

*Source: _072612a - Annex I Table 2: Demand for Fund Financing in Response to Policy and/or External Shocks (Decadal-change distributions; sample of 75 LICs)._

### The Impact of Short-Term Fund Engagement on Macroeconomic Outcomes (PSM estimates)

### Sample and coverage
- Sample size: 58 LICs.
- Time coverage: 1980-2010.
- Estimation approach: Propensity-score matching (PSM). PS stands for the propensity score indicating the likelihood of Fund programs addressing immediate balance of payments needs.
- Outcome measure: Estimated impact of short-term Fund engagement on changes in macroeconomic outcomes relative to the control group having similar propensity scores.
- Subsamples reported: All, PS>0.5, PS>0.7.
- Changes in macroeconomic outcomes refer to first differences of the outcome variables.

### Outcomes and reported axes/metrics (Figures 4 and 5)
- Real GDP Growth (In percent) — axis example: 0.0 to 2.5.
- Inflation (In percent) — axis example: -25.0 to 0.0 and -14.0 to 0.0 shown across figures.
- Reserve coverage (In months of imports) — axis example: 0.0 to 0.9.
- Government balance (In percent of GDP) — axis example: 0.0 to 2.5 and 0.0 to 1.0 across figures.
- Current account balance plus FDI (In percent of GDP) — axis example: 0.0 to 3.5 and 0.0 to 4.0 across figures.
- Change in REER (In percent) — axis example: -6.0 to 0.0.
- Real health spending per capita (Percent change) — axis example: 0.0 to 16.0.
- Real education spending per capita (Percent change) — axis example: -4.5 to 1.0.

### Interpretive notes from figures
- Figures show estimated impacts of short-term Fund engagement (Fund-supported programs minus the control group) for the full sample and for subsamples defined by propensity score thresholds (PS>0.5 and PS>0.7).
- Source annotation for these figures: Staff estimates.
- Notes reiterate that changes reported are first differences and that the sample is composed of 58 LICs covering 1980-2010.

*Source: _072612a - Annex I Table 2: Demand for Fund Financing in Response to Policy and/or External Shocks (Figures 4 and 5; PSM impact estimates; sample of 58 LICs)._

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2012/_072612a.pdf_