## _wp13273 ## Source details **Canonical URL:** [_wp13273](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13273.pdf) ## Other formats - [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13273.pdf.md) - [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13273.pdf.json) --- ### Major themes and chapter structure - Chapter 1: "Summary of Literature on the Impact of IMF Programs, 2000–12" (page 11) - Chapter 2: "Impact of Longer-Term IMF Engagement on Economic Performance" (page 18) - Chapter 3: "Determinants of Long-Term Real Per Capita GDP Growth" (page 21) - Chapter 4: "Results: Impact of Short-Term IMF Engagement by Propensity Score Matching" (page 28) - Chapter 5: "Determinants of Longer-Term IMF Engagement" (page 35) - Chapter 6: "Demand for IMF Financing in Response to Policy and/or External Shocks" (page 38) - Chapter 7: "Impact of Longer-Term IMF Engagement on Economic Performance Using Pre-2000s and Post-2000s Periods" (page 49) - Chapter 8: "Impact of Longer-Term IMF Engagement on Economic Performance Using Non-Overlapping Periods" (page 50) - Chapter 9: "Impact of Longer-Term IMF Engagement on Economic Performance With No Adjustment for Program Implementation" (page 51) - Chapter 10: "Impact of Short-Term IMF Engagement With No Adjustment for Implementation" (page 52) - Chapter 11: "Impact of Short-Term IMF Engagement, 1980–99" (page 53) - Chapter 12: "Impact of Short-Term IMF Engagement, Matching on Propensity Score and ODA Disbursements" (page 54) - Chapter 13: "Impact of Short-Term IMF Engagement, Matching on Propensity Score and Lagged GDP Growth" (page 55) - Chapter 14: "Short-Term IMF Engagement: Rosenbaum Sensitivity Analysis to Hidden Selection Bias" (page 56) ### Methodology and data - Unit of analysis: decadal averages with 50 percent overlap; periods: 1986–95; 1991–00; 1996–05; 2001–10. - Longer-term IMF engagement dummy: equals 1 if a country had five or more years of IMF-supported programs in the 10-year period; a year qualifies if a country had a SAF, ESAF, PRGF, ECF, SBA, ESF-HAC, SCF, or PSI for at least six months in that year. - Program-years purged of episodes with prolonged program interruptions defined as a delay of more than six months in completing a review owing to noncompliance with macroeconomic performance criteria. - Primary identification strategies: - Propensity Score Matching (PSM) using pooled panel probit selection equation. - Panel regressions on 10-year period averages using dynamic panel GMM (one-step system GMM with robust standard errors) with inverse Mills Ratio to control for selection. - PSM matching algorithms: nearest-neighbor, five-nearest-neighbor, radius matching, Kernel matching; all estimates based on common support. - Sample: analyses use a panel dataset of 75 LICs for longer-term engagement analyses; short-term engagement sample composed of 58 LICs covering 1980–2010. ### Key stylized facts (LICs, decadal comparisons) - Over last two decades LICs on average experienced increases in: - real GDP per capita growth, - government balances, - reserves, - current account balances, - foreign direct investment (FDI), - exports, - institutional quality, - social spending. - Concurrent reductions in: - economic volatility, - inflation, - external debt, - poverty. - LICs with longer-term IMF engagement generally started from weaker 1980s conditions but saw larger increases in real GDP per capita growth, government balance, exports, FDI, and social spending, and larger reductions in volatility, inflation, and external debt. ### Main empirical findings — Longer-term IMF engagement (PSM and panel results) - PSM average treatment-on-treated (ATT) estimates (selected, coefficients with bootstrapped standard errors): - GDP per capita growth: - Nearest-Neighbor 3.38*** (0.80) - Five-Nearest-Neighbor 3.41*** (0.73) - Radius 3.30*** (0.75) - Kernel 3.50*** (0.68) - GDP per capita growth volatility: - Nearest-Neighbor -1.85 (1.33) - Five-Nearest-Neighbor -2.33** (0.99) - Radius -2.02*** (0.75) - Kernel -2.45*** (0.90) - Inflation: - Nearest-Neighbor -10.57* (5.60) - Five-Nearest-Neighbor -10.46** (4.21) - Radius -10.75*** (3.98) - Kernel -9.60** (4.22) - Government balance: - Nearest-Neighbor 3.58*** (1.16) - Five-Nearest-Neighbor 3.52*** (1.02) - Radius 1.59 (1.15) - Kernel 3.38*** (0.95) - FDI: - Nearest-Neighbor 1.76*** (0.39) - Five-Nearest-Neighbor 1.86*** (0.39) - Radius 1.06*** (0.38) - Kernel 1.85*** (0.35) - Education spending: - Nearest-Neighbor 0.78 (0.58) - Five-Nearest-Neighbor 0.80* (0.44) - Radius 1.01*** (0.27) - Kernel 0.80* (0.49) - Social spending: - Nearest-Neighbor 1.07 (0.67) - Five-Nearest-Neighbor 1.00* (0.52) - Radius 1.06*** (0.34) - Kernel 0.91 (0.57) - Poverty gap: - Nearest-Neighbor -3.90 (3.00) - Five-Nearest-Neighbor -3.98* (2.39) - Radius -3.83** (1.89) - Kernel -3.89 (2.59) - Gini: - Nearest-Neighbor -4.94* (2.56) - Five-Nearest-Neighbor -3.67* (2.14) - Radius -3.84*** (1.38) - Kernel -3.58* (2.09) - Reserve coverage, Tax revenue, External debt, Aid: mixed significance across methods (see reported coefficients and standard errors). - Significance notation: * 10 percent; ** 5 percent; *** 1 percent. - Panel growth regression (system GMM) — longer-term engagement dummy: - Longer-term Fund engagement coefficients reported: 3.52* (2.08); 3.35 (2.13); 3.18* (1.92); 3.53* (2.15) across robustness columns. - Inclusion of inflation and growth volatility reduces magnitude/significance of IMF dummy, indicating these are likely transmission channels. - Controlling for both longer-term engagement and net IMF disbursements: only the longer-term engagement dummy remains significant, suggesting policy advice and capacity building rather than scale of financing drive long-term growth effects. - Education coefficient consistently positive and significant: 0.14** (0.06); 0.13** (0.05); 0.13** (0.06); 0.14** (0.07) in reported specifications. - Observations/countries: 181/54; 178/54; 181/54; 181/54 depending on specification. - Diagnostic tests: Arellano-Bond AR(1): 0.16; 0.29; 0.20; 0.20. AR(2): 0.56; 0.36; 0.43; 0.56. Sargan: 1.00; 0.99; 0.99; 1.00. Hansen: 0.38; 0.41; 0.29; 0.34. ### Robustness of longer-term results - Alternative period splits (pre-2000s vs. post-2000s) yield qualitatively similar results; some differences in significance (e.g., reserve coverage positive and significant pre-2000s; government balance not significant pre-2000s; post-2000s weaker). - Non-overlapping period estimations remain qualitatively similar. - Not adjusting for program implementation produces larger coefficients; adjusting for implementation yields smaller coefficients — presented estimates likely a lower bound. - Rosenbaum sensitivity: unobserved country characteristics would need to increase the odds ratio of being in longer-term engagement by more than 300 percent before baseline results would be biased (reported as well above 100 percent rule-of-thumb). ### Main empirical findings — Short-term IMF engagement (PSM results, levels and changes) - Short-term engagement targeted at immediate balance of payments needs (sample 58 LICs; 1980–2010). - Impacts on levels (selected): - Growth: - Full sample: growth estimated 0.9 percent higher than control group (aggregate narrative); table entries report variants (e.g., Real GDP growth (%): 0.53; 0.87; 1.33** with std. errors (0.49)(0.56)(0.62) in Table 10). - High propensity groups: impact rises to 1¼–1¾ percent and becomes significant only for high propensity scores. - Inflation (level): - All LICs: -12.09*** (4.11) in one specification; Table 10 reports -13.35*** (3.92); other subgroup magnitudes larger (e.g., PS>0.7: -15.83*** (5.76)). - Reserve coverage (months of imports, level): - All LICs: 0.77*** (0.15) in one table; Table 10 reports 0.58*** (0.13) with subgroup larger values (e.g., LICs with weaker fundamentals: 0.99*** (0.16)). - Current account balance plus FDI (% of GDP, level): - All LICs: 2.26** (0.98) in one summary table; Table 10 reports 2.14** (0.86). - Government balance (% of GDP, level): - All LICs: 2.23*** (0.63) in one summary table; Table 10 reports 1.64*** (0.58) in the no-adjustment specification. - ODA commitments (% of GDP, level): - All LICs: 2.19*** (0.61) in summary; Table 10 reports 1.88*** (0.53) in another specification. - Impacts on changes (X(t)-X(t-1), selected): - Change in real GDP growth (%): - All LICs: 0.74 (0.75); PS>0.7: 2.21** (1) - Change in inflation (%): - All LICs: -6.45 (4.6); LICs with weaker fundamentals: -10.47* (5.33); PS>0.7: -13.96** (6.22) - Change in reserve coverage (months): - All LICs: 0.71*** (0.1); LICs with weaker fundamentals: 0.93*** (0.11); PS>0.7: 0.79*** (0.12) - Change in current account + FDI (% of GDP): - All LICs: 1.44** (0.71); LICs with weaker fundamentals: 2.03** (0.84); PS>0.7: 2.43** (0.97) - Change in government balance (% of GDP): - All LICs: 0.78* (0.46); LICs with weaker fundamentals: 1.06* (0.55); PS>0.7: 1.41** (0.59) - Short-term engagement results robust to multiple sensitivity checks: - Not adjusting for implementation yields qualitatively similar impacts but generally smaller magnitudes in some specifications. - 1980–1999 sample: qualitatively similar, sometimes quantitatively stronger impacts. - Conditioning matching on ODA disbursements: levels similar; impacts on growth and government balances become insignificant after controlling for ODA disbursements in some specifications, consistent with a catalytic role of IMF programs on ODA. - Conditioning on lagged GDP growth: lagged growth raises likelihood of program request; conditioning yields qualitatively similar results and often strengthens growth impact significance. ### Determinants of program participation (selection equations) - Determinants of longer-term engagement (pooled probit; Table 5 coefficients): - Initial reserves -0.156*** (0.04) - Initial aid/GDP 0.016* (0.01) - Trading partner growth -0.092 (0.07) - IMF quota/GDP -0.078** (0.03) - Resource rents/GDP -0.020** (0.01) - Landlockedness 0.741*** (0.21) - Political connectedness 0.010* (0.01) - Polity 0.010 (0.02) - Constant 0.362 (0.52) - Observations 203 - Interpretation: higher initial reserves reduce propensity for longer-term engagement; landlocked and resource-poor countries have higher propensity; larger quota and lower political connectedness imply lower probability. - Determinants of short-term demand for IMF financing (correlated random effects probit; Bal Gündüz (2009), Table 6 coefficients, t-statistics): - Current account balance to GDP (t-1) -0.076*** (-4.61) - Reserve coverage (months of imports) (t-1): CFA -0.478*** (-6.08); non-CFA -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) - Paris Club dummy 0.774*** (3.24) - Country-specific averages: Total debt service to exports 0.044*** (2.63); FDI to GDP -0.105* (-1.76) - Pseudo R2 0.58; Observations 532; Countries 55; Sample probability 0.44 - Interpretation: lower reserves, worse current account, lower growth, adverse terms of trade, and some global shocks raise probability of short-term IMF financing. ### Mechanisms and interpretation - Longer-term engagement channels: - Macroeconomic stabilization (reductions in inflation and growth volatility). - Institutional development and capacity building (CPIA improvements in some specifications). - Shifts in spending composition favoring health and education and improved targeting. - Catalytic and policy-advice roles rather than scale of IMF financing: panel regressions show net IMF disbursements not significant when longer-term engagement dummy included. - Short-term engagement channels: - IMF financing eases short-term adjustment, raises reserves and current account plus FDI, lowers inflation and fiscal deficits. - Potential catalytic effects on ODA: commitments and disbursements of ODA higher for program group; differences larger for commitments than disbursements. - Reserves and IMF role: - IMF presence can act as partial insurance reducing incentive to accumulate reserves; expected reserve dynamics depend on nature of shocks and program objectives. - Poverty and distribution: - Longer-term engagement associated with larger declines in poverty gap and reductions in Gini; poverty rate declines larger but not always statistically significant due to limited data. ### Robustness and sensitivity analyses - Sensitivity to hidden bias (Rosenbaum bounds and Hodges-Lehmann): - Hodges-Lehmann point estimates indicate unobserved characteristics would have to increase odds ratio of longer-term engagement by more than 300 percent to overturn baseline results. - Rosenbaum Γ reported in Table 14 for selected outcomes (All LICs vs LICs with Weaker Fundamentals); examples: - Real GDP growth All LICs: Γ 1.38 Probability 0.044; LICs with Weaker Fundamentals: Γ 1.76 Probability 0.046. - Inflation All LICs: Γ 1.14 Probability 0.045; LICs with Weaker Fundamentals: Γ 1.27 Probability 0.048. - Reserve coverage All LICs: Γ 2.11 Probability 0.046; LICs with Weaker Fundamentals: Γ 2.57 Probability 0.049. - Interpretation: some variables (e.g., reserve coverage) relatively robust to hidden bias; others (e.g., inflation) more sensitive. - Other robustness checks: - Alternative period splits (pre-2000s/post-2000s), non-overlapping periods, inclusion/exclusion of program-implementation adjustments, conditioning on ODA disbursements or lagged GDP growth — results remain broadly qualitatively similar with some quantitative variation. ### Policy implications and recommendations - Dual role of IMF support for LICs: - Longer-term policy support (ECF and predecessors; PSI) is associated with higher long-term growth, reduced volatility, lower inflation, higher government balances, higher social spending, higher FDI, and reductions in poverty and inequality. These effects appear linked to IMF policy advice and capacity building rather than the scale of financing. - Short-term liquidity support (augmentations, SCF, RCF, emergency instruments) is positively associated with immediate stabilization outcomes: higher short-term growth, higher reserve coverage, improved current account balances, and lower inflation and fiscal deficits—especially for countries with pronounced balance of payments needs. - Operational recommendations: - Maintain a diverse set of IMF instruments for LICs: medium-term policy support (ECF, PSI) for structural and capacity objectives and quick short-term financing (ECF augmentations, SCF, RCF) for urgent balance of payments needs. - Use precautionary arrangements (ECF, SCF) as insurance in absence of shocks to provide policy support even at low access levels. - Account for program implementation: realistic assessment and monitoring of implementation can affect estimated impacts and should inform program design. - Recognize potential catalytic role on ODA: improve predictability and utilization of donor flows in program contexts to enhance effectiveness. *Source: IMF staff calculations and analysis as presented in _wp13273.* ### 1. Summary of Literature on the Impact of IMF Programs, 2000–12 .....................................11 ### _wp13273 - 1. Summary of Literature on the Impact of IMF Programs, 2000–12 .....................................11 ### Major themes and chapter structure - Chapter 1: "Summary of Literature on the Impact of IMF Programs, 2000–12" (page 11) - Chapter 2: "Impact of Longer-Term IMF Engagement on Economic Performance" (page 18) - Chapter 3: "Determinants of Long-Term Real Per Capita GDP Growth" (page 21) - Chapter 4: "Results: Impact of Short-Term IMF Engagement by Propensity Score Matching" (page 28) - Chapter 5: "Determinants of Longer-Term IMF Engagement" (page 35) - Chapter 6: "Demand for IMF Financing in Response to Policy and/or External Shocks" (page 38) - Chapter 7: "Impact of Longer-Term IMF Engagement on Economic Performance Using Pre-2000s and Post-2000s Periods" (page 49) - Chapter 8: "Impact of Longer-Term IMF Engagement on Economic Performance Using Non-Overlapping Periods" (page 50) - Chapter 9: "Impact of Longer-Term IMF Engagement on Economic Performance With No Adjustment for Program Implementation" (page 51) - Chapter 10: "Impact of Short-Term IMF Engagement With No Adjustment for Implementation" (page 52) - Chapter 11: "Impact of Short-Term IMF Engagement, 1980–99" (page 53) - Chapter 12: "Impact of Short-Term IMF Engagement, Matching on Propensity Score and ODA Disbursements" (page 54) - Chapter 13: "Impact of Short-Term IMF Engagement, Matching on Propensity Score and Lagged GDP Growth" (page 55) - Chapter 14: "Short-Term IMF Engagement: Rosenbaum Sensitivity Analysis to Hidden Selection Bias" (page 56) ### Methodology and analytical tools (as organized in the content) - Propensity score matching (PSM) methodology is presented (Annex I; page 31) - Panel regression on the determinants of long-term growth is presented (Annex II; page 40) - Results include both short-term and longer-term engagement analyses, with sensitivity and robustness checks: - Propensity Score Matching approaches (multiple chapters) - Rosenbaum sensitivity analysis for hidden selection bias (Chapter 14; page 56) - Comparisons across time periods: pre-2000s vs post-2000s, non-overlapping periods (Chapters 7 and 8) ### Empirical focus and outcomes examined - Economic performance outcomes tied to IMF engagement: - Long-term real per capita GDP growth (Chapter 3; page 21) - Macroeconomic outcomes for short-term IMF engagement (Chapter 4; page 28; Chapters 10–13) - Demand for IMF financing in response to policy and/or external shocks (Chapter 6; page 38) - Implementation adjustments and their role in estimated impacts: - Analyses with and without adjustment for program implementation (Chapters 9 and 10; pages 51–52) ### Figures and empirical illustrations - Figures listed and their focus: - Figure 1: "Years Spent by LICs Under IMF-Supported Programs, 1986–2011" (page 5) - Figure 2: "Changes in Average Decadal GDP Per Capita Growth and Poverty Gaps, 1986–2010" (page 14) - Figure 3: "Macroeconomic Conditions in LICs Across Decades" (page 41) - Figure 4: "Macroeconomic Conditions in LICs Across Decades and County Groupings" (page 43) - Figure 5: "Changes in Macroeconomic Performance of LICs" (page 44) - Figure 6: "The Impact of Short-Term IMF Engagement on Macroeconomic Outcomes" (page 47) - Figure 7: "The Impact of Short-Term IMF Engagement on Changes in Macroeconomic Outcomes" (page 48) ### Scope of robustness checks and temporal coverage - Temporal coverage spans multiple decades with explicit attention to: - 1980–99 analyses for short-term engagement (Chapter 11; page 53) - 1986–2011 series for program exposure in low-income countries (Figure 1; page 5) - Decadal comparisons including 1986–2010 (Figure 2; page 14) - Robustness approaches include: - Matching on additional covariates (ODA disbursements; lagged GDP growth) (Chapters 12–13; pages 54–55) - Sensitivity to hidden selection bias (Chapter 14; page 56) *Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13273.pdf* ### References ............................................................................................................. ### _wp13273 - References ............................................................................................................. ### Abbreviations and acronyms - CFFs Compensatory Financing Facilities - CPIA Country Policy and Institutional Assessment - ECF Extended Credit Facility - EFFs Extended Fund Facilities - ENDA Emergency Natural Disaster Assistance - EPCA Emergency Post-Conflict Assistance - ESAF Enhanced Structural Adjustment Facility - ESF Exogenous Shocks Facility - FDI Foreign Direct Investment - GEE General Evaluation Estimator - GMM Generalized Method of Moments - HAC High Access Component - HIPC Heavily Indebted Poor Countries - IEO Independent Evaluation Office - LICs Low-Income Countries - MDRI Multilateral Debt Relief Initiative - MICs Middle-Income Countries - ODA Official Development Assistance - PITF Political Instability Task Force - PRGF Poverty Reduction and Growth Facility - PSI Policy Support Instrument - PSM Propensity Score Matching - RCFs Rapid Credit Facilities - SAF Structural Adjustment Facility - SBAs Stand-By Arrangements - SCF Standby Credit Facility - SMP Staff-Monitored Program ### I. Overview — scope and key premises - The last 25 years have seen marked improvements in macroeconomic policies and long-term increases in real GDP per capita growth together with reductions in poverty in LICs. - The paper assesses how LIC involvement in IMF-supported programs may have affected economic developments in LICs over the past quarter century. - During this period, the IMF has engaged in financial or non-financial arrangements with more than half of all LICs, and more than three-quarters of all IMF-supported programs have been with LIC members. - Major methodological challenge: selection bias — countries approach the IMF often because they are already facing economic difficulties, making simple comparisons with non-program countries potentially misleading. - Distinct characteristics of LICs (versus emerging markets) emphasized: - (i) Nature of shocks: LICs are more vulnerable to frequent domestic and external shocks (terms of trade shocks, demand shocks, natural disasters, domestic or regional instability). - (ii) Access to financing: LICs have much less access to domestic or external financing, making them more dependent on donor assistance and IMF-supported programs. - (iii) Longer-term challenges: IMF programs with LICs emphasize capacity and institution building and medium- and longer-term objectives important for poverty reduction and growth. - (iv) Engagement pattern: IMF engagement with most LICs has been less episodic and more continuous. - Program-years definition (Figure 1 notes): years when a country had a SAF, ESAF, PRGF, ECF or PSI for at least six months. Sample composed of 75 out of the 78 LICs as of January 2010; Somalia, Timor-Leste, and Tonga had no available data. - This study breaks IMF-supported programs with LICs into two subsets: prolonged support and short-term shock response, and applies methodological refinements including Propensity Score Matching (PSM) and consideration of program implementation. ### Key findings from overview and summary of results - Evidence suggests longer-term IMF program support may have helped LICs sustain economic growth and boost resilience by building fiscal buffers (Section III). - Between 1986 and 2010, controlling for selection bias, LICs with IMF-supported programs experienced, on average, significantly higher: - real per capita GDP growth, - fiscal balances, - foreign investment, - social spending, compared to LICs without such programs. - LICs with longer-term IMF engagement tended to attain significant reductions in: - poverty, - income inequality, - inflation, - growth volatility, relative to their control group. - Controlling for the presence of longer-term IMF engagement, the scale of IMF financing does not appear to be significant in determining economic growth over long time frames — suggesting role of IMF as policy advisor and capacity builder may dominate lending role for long-term growth. - IMF financial support shows greatest impact when LICs face substantial short-term macroeconomic imbalances or exogenous shocks: stepped-up IMF financing through augmentations of existing programs or short-term and emergency facilities is positively associated with short-term growth and indicators of macroeconomic stability (Section IV). ### II. Overview of the empirical literature and contribution of the current study - Main literature challenge: endogeneity/self-selection bias in participation in IMF-supported programs; countries requesting IMF arrangements often do so when in need. - Participation determinants: early research emphasized economic determinants; later work incorporated political variables to capture “supply” side of programs. Results are mixed and models have low predictive power for concessional programs. - Few studies focus exclusively on LICs or on concessional programs; repeated or prolonged use of Fund resources is a phenomenon more peculiar to LICs. - Consensus in broader literature: - IMF-supported programs are associated with significant improvement in the balance of payments and some effect on inflation. - Evidence on impact on growth is mixed. - Methodological notes from literature survey: - Many prior studies examined non-concessional programs (SBAs, EFFs) on mixed samples; few isolated LICs or concessional programs. - Few studies explore the impact of prolonged engagement on longer-term growth; Independent Evaluation Office (IEO, 2002) found negative effects on growth for some prolonged users but not for those under concessional arrangements. - Most studies do not sufficiently account for program implementation/compliance, despite its importance. - Correcting for selection bias has become standard more recently: Heckman two-stage, instrumental variables (IV), and Propensity Score Matching (PSM) have been used. Identifying valid exclusion restrictions remains a key challenge. - Microeconometric impact-evaluation techniques (matching) have been applied in a subset of recent research; results vary by outcome and sample. - Table 1 (literature summary) synthesizes studies 2000–12: varied findings across GDP growth, inflation, fiscal deficit, current account balance, monetary growth, Gini coefficient, education spending, health spending, poverty; notation used in table: +* Significantly positive; -* Significantly negative; + Positive but insignificant; - Negative but insignificant; 0 Very close to zero. - Contribution of current paper (four points): - Focus solely on LICs. - Disaggregation into longer-term prolonged support versus short-term episodic support. - Use of PSM over a sample covering nearly three decades ending in 2010, examining a wide range of macroeconomic and social outcomes. - Investigation of transmission channels distinguishing IMF financing versus policy advice and capacity development. ### III. Impact of longer-term IMF engagement in LICs — stylized facts and econometric approach - Stylized facts (decadal comparison and distributions): - Over last two decades, LICs on average experienced increases in: - real GDP per capita growth, - government balances, - reserves, - current account balances, - foreign direct investment (FDI), - exports, - institutional quality, - social spending. - Simultaneously LICs recorded reductions in: - economic volatility, - inflation, - external debt, - poverty. - These improvements hold across country sizes, geographical groupings, institutional capacity (CPIA), and per-capita income. - LICs with longer-term IMF program engagement (extensive program engagement) experienced, on average: - Comparatively stronger improvement in long-term economic performance versus LICs without extensive engagement. - Faced comparatively weaker initial economic conditions in the 1980s but experienced larger increases in: - real GDP per capita growth, - government balance, - exports, - FDI, - social spending. - More marked reduction in economic volatility, inflation, and external debt. - This improvement largely closed the performance gap that existed relative to other LICs around the time ESAF was created in 1987. - Econometric analysis — questions and methods: - Primary questions: a. How does longer-term IMF program engagement affect macroeconomic performance (including growth and institutional variables)? — approach: Propensity Score Matching (PSM). b. What is the longer-term impact on economic growth and associated transmission channels? — approach: panel regressions on 10-year period averages with controls for traditional determinants of long-run growth and a dummy for longer-term IMF engagement. - PSM approach: two-stage process: (i) First-stage regression estimates propensity score (probability) of a country becoming a longer-term user of IMF-supported programs. (ii) Compare average economic performance over a 10-year period between longer-term program users and others with similar propensity scores. - Panel regressions: use 10-year period averages to identify channels — macroeconomic stabilization, institutional development, and provision of development financing. - Data and sample notes: - Analysis uses a panel dataset of 75 LICs and decadal averages spanning the period 1986– (text truncated at provided content). *Source: IMF staff calculations and analysis as presented in the provided content unit.* ### 2010. Given the focus on longer-term engagement we work with decadal averages where ### _wp13273 - 2010. Given the focus on longer-term engagement we work with decadal averages where ### Methodology and data construction - Decadal averages are the unit of analysis; 10-year periods share a 50 percent overlap. The periods used are: 1986–95; 1991–00; 1996–05; 2001–10. - Longer-term IMF engagement dummy: equals 1 if a country had five or more years of IMF-supported programs in the 10-year period and zero otherwise. A country is considered to have longer-term engagement in a given decade if in five or more years it had a financial arrangement or a Policy Support Instrument in place, for at least six months in each of these years. - Qualifying IMF arrangements: ECF and predecessors (PRGF, ESAF, SAF), SBA, ESF-HAC, SCF, and the Policy Support Instrument (PSI). - Program years were purged of episodes with prolonged program interruptions: defined as a delay of more than six months in completing a review owing to noncompliance with macroeconomic performance criteria. - Propensity Score Matching (PSM): pooled panel probit selection equation (to avoid incidental parameters problem) with independent variables including initial macroeconomic buffers (reserve coverage ratio; foreign aid to GDP ratio at the beginning of each decade), structural and institutional characteristics (landlocked dummy, political connectedness, natural resource rents, CPIA), trading partners’ real GDP growth, and IMF quota. - PSM estimations run using four matching approaches: nearest-neighbor matching, five-nearest neighbor matching, radius matching, and Kernel matching. All estimates are based on matches within the “common support” sample. - Panel growth regressions: dynamic panel GMM (one-step system GMM with robust standard errors). Endogeneity of longer-term engagement controlled via the inverse Mills Ratio from the first-stage selection equation. Variables added in augmented specifications are those significant in PSM and common determinants of growth. ### Main PSM findings (impacts of longer-term IMF engagement) - Longer-term IMF engagement leads to significantly higher long-term real per capita GDP growth compared to initial conditions. - Longer-term IMF users exhibit significantly higher reductions in growth volatility and inflation; improvements in the government balance are larger for longer-term users. - Longer-term IMF engagement is associated with significantly larger increases in FDI. - Changes in social spending, particularly education spending, are larger for countries with longer-term IMF engagement; health spending changes are positive but not statistically significant. - Poverty gap decreased more for countries with longer-term IMF engagement; declines in poverty rates are larger but not statistically significant (poverty data limited, especially earlier years). - Longer-term IMF engagement is associated with significantly greater reductions in income inequality (Gini); coefficients consistently significant across matching techniques despite limited historical data. - Changes in reserve coverage, tax revenue, and CPIA are larger for longer-term users but generally not significantly different from controls (exception: CPIA under one estimation). - Relationship between longer-term engagement and changes in aid (catalytic effect), external debt, and the current account is inconclusive across the four estimation techniques. ### Selected PSM estimates (Table 2) — coefficient estimates and bootstrapped standard errors (parentheses) - GDP per capita growth: Nearest-Neighbor 3.38*** (0.80); Five-Nearest-Neighbor 3.41*** (0.73); Radius 3.30*** (0.75); Kernel 3.50*** (0.68). - GDP per capita growth volatility: Nearest-Neighbor -1.85 (1.33); Five-Nearest-Neighbor -2.33** (0.99); Radius -2.02*** (0.75); Kernel -2.45*** (0.90). - Inflation: Nearest-Neighbor -10.57* (5.60); Five-Nearest-Neighbor -10.46** (4.21); Radius -10.75*** (3.98); Kernel -9.60** (4.22). - Government balance: Nearest-Neighbor 3.58*** (1.16); Five-Nearest-Neighbor 3.52*** (1.02); Radius 1.59 (1.15); Kernel 3.38*** (0.95). - FDI: Nearest-Neighbor 1.76*** (0.39); Five-Nearest-Neighbor 1.86*** (0.39); Radius 1.06*** (0.38); Kernel 1.85*** (0.35). - Education spending: Nearest-Neighbor 0.78 (0.58); Five-Nearest-Neighbor 0.80* (0.44); Radius 1.01*** (0.27); Kernel 0.80* (0.49). - Social spending: Nearest-Neighbor 1.07 (0.67); Five-Nearest-Neighbor 1.00* (0.52); Radius 1.06*** (0.34); Kernel 0.91 (0.57). - Poverty gap: Nearest-Neighbor -3.90 (3.00); Five-Nearest-Neighbor -3.98* (2.39); Radius -3.83** (1.89); Kernel -3.89 (2.59). - Poverty rate: Nearest-Neighbor -2.16 (3.18); Five-Nearest-Neighbor -2.98 (2.66); Radius -2.51 (2.43); Kernel -2.49 (2.81). - Gini: Nearest-Neighbor -4.94* (2.56); Five-Nearest-Neighbor -3.67* (2.14); Radius -3.84*** (1.38); Kernel -3.58* (2.09). - Reserve coverage: Nearest-Neighbor 0.18 (0.45); Five-Nearest-Neighbor 0.20 (0.35); Radius 0.32 (0.30); Kernel 0.15 (0.33). - Tax revenue: Nearest-Neighbor 0.47 (0.90); Five-Nearest-Neighbor 0.92 (0.73); Radius 1.23* (0.69); Kernel 0.93 (0.76). - External debt: Nearest-Neighbor 50.23 (38.77); Five-Nearest-Neighbor -17.53 (27.18); Radius -4.19 (22.63); Kernel -25.65 (19.90). - Aid: Nearest-Neighbor -0.45 (1.72); Five-Nearest-Neighbor 0.17 (1.45); Radius -0.26 (1.45); Kernel -0.19 (1.32). (Note: Bootstrapped standard errors in parentheses. *10 percent significance; **5 percent significance; ***1 percent significance. CPIA = Country Policy and Institutional Assessment; FDI = Foreign Direct Investment.) ### Panel growth regressions and transmission channels (Table 3 highlights) - Longer-term Fund engagement coefficients (robustness columns): 3.52* (2.08); 3.35 (2.13); 3.18* (1.92); 3.53* (2.15). - Controlling for inflation and growth volatility reduces the IMF dummy coefficient, indicating that declines in inflation and growth volatility are likely transmission channels for the IMF’s longer-term impact on per capita GDP growth. - When controlling for both the longer-term IMF engagement dummy and the size of net IMF disbursements in the decade, only the longer-term IMF engagement dummy is significant — suggesting IMF policy support matters for longer-term growth performance rather than the overall level of financing. - Other reported coefficients (selected): Initial GDP per capita 0.97 (2.42); 0.79 (2.57); 0.49 (2.19); 0.92 (2.75). Education 0.14** (0.06); 0.13** (0.05); 0.13** (0.06); 0.14** (0.07). - Observations/countries: 181/54; 178/54; 181/54; 181/54. - Arellano-Bond AR(1) in first differences: 0.16; 0.29; 0.20; 0.20. AR(2): 0.56; 0.36; 0.43; 0.56. Sargan test of overidentifying restrictions: 1.00; 0.99; 0.99; 1.00. Hansen test of overidentifying restrictions: 0.38; 0.41; 0.29; 0.34. ### Robustness checks and sensitivity - Alternative period splits (pre-2000s vs. post-2000s) yield qualitatively similar results. Differences: effect on reserve coverage positive and significant in pre-2000s; government balance effect not statistically significant in pre-2000s; post-2000s results statistically weaker; insufficient poverty data for pre-2000s. - Re-estimations with non-overlapping observations (smaller sample) remain qualitatively similar. - Alternative participation equation including a lagged dummy of longer-term IMF engagement: lagged dummy significant, but second-stage regressions remain qualitatively similar; authors present results without lagged dummy. - Adjusting for program implementation (purging years with interruptions > six months) yields smaller coefficients than when not adjusted; coefficients tend to be larger when program implementation is not accounted for, suggesting the presented coefficients may be a lower bound of the true impact. - Including foreign aid in panel regressions does not substantially change results; aid not statistically significant in the growth regressions reported. - Sensitivity to unobserved heterogeneity (Rosenbaum bounds method): Hodges-Lehmann point estimates indicate unobserved country characteristics would have to increase the odds ratio of being in a longer-term engagement with the IMF by more than 300 percent before they would bias the estimated baseline results (well above the 100 percent rule-of-thumb). ### Interpretation of mechanisms - Potential channels through which longer-term IMF engagement affects outcomes include: - Fiscal targets creating fiscal space and changes in spending composition favoring health and education or better targeting the poorest and most vulnerable. - IMF assistance in designing consistent macroeconomic frameworks and providing regular independent policy advice. - IMF role in debt relief initiatives. - IMF-provided capacity building via policy support and technical assistance. - For reserves: presence of the IMF may act as partial insurance, reducing incentives to accumulate reserves; LICs often face protracted balance of payments needs where Fund programs may not aim to boost reserves. - For catalytic effects on aid and debt: averaging over a decade may mask short-run catalytic effects at program start or when debt relief is granted; debt dynamics over a decade can reflect multiple events (debt relief, concessional financing, scaling up of capital projects) that complicate decade-average measurement. *Source: IMF staff calculations from the cited study.* ### conclusion that programs “cause” these crises along with adverse effects on macroeconomic ### _wp13273 - conclusion that programs “cause” these crises along with adverse effects on macroeconomic ### Methodology: Propensity Score Matching (PSM) - First stage: annual probability of participating in IMF-supported programs estimated conditional on observable economic conditions and country characteristics. - Second stage: use propensity scores to match program countries to non-program countries to construct a control group (see Annex I for details). - Sample period covers 1980–2010 and is composed of 58 low-income countries (LICs). - The set of arrangements includes those addressing an immediate balance of payments need arising from policy and/or exogenous shocks: SBAs, SAF/ESAF/PRGF/ECF augmentations, Compensatory Financing Facilities (CFFs), ESFs, SCFs, and Rapid Credit Facilities (RCFs). ### Determinants of Program Participation - Higher likelihood of IMF financing associated with: - Lower reserve coverage. - Deterioration in the current account balance. - Weaker real GDP growth. - Increased macroeconomic instability: higher fiscal deficits, inflation, and exchange market pressures. - Adverse terms of trade shocks. - Global conditions matter: changes in real oil and non-oil commodity prices and world trade are significant determinants and could create cycles in demand for IMF financing. - Persistent differences in debt service burden and resource inflows among LICs are associated with unobserved country heterogeneity. ### Empirical findings: Short-term IMF engagement (from Table 4 and narrative) - Aggregate and subgroup impacts (All LICs; LICs with weaker fundamentals; PS>0.5; PS>0.7): - Growth: - Full sample: growth is estimated to be 0.9 percent higher than the control group. - High propensity group: impact rises to 1¼–1¾ percent and becomes significant only for countries with high propensity scores. - Change in growth: positive and significant only for those with high propensity scores. - Inflation (level): - All LICs: -12.09*** (standard error 4.11) - LICs with weaker fundamentals: -14.30*** (standard error 4.95) - PS>0.7: -15.83*** (standard error 5.76) - Reserve coverage (in months of imports, level): - All LICs: 0.77*** (0.15) - LICs with weaker fundamentals: 0.99*** (0.16) - PS>0.7: 0.81*** (0.16) - Current account balance plus FDI (% of GDP, level): - All LICs: 2.26** (0.98) - LICs with weaker fundamentals: 2.98*** (1.14) - PS>0.7: 3.97*** (1.32) - Government balance (% of GDP, level): - All LICs: 2.23*** (0.63) - LICs with weaker fundamentals: 2.50*** (0.77) - PS>0.7: 2.89*** (0.88) - Change in real health spending per capita (%): - All LICs: 8.38* (5.02) - LICs with weaker fundamentals: 8.59 (6.74) - PS>0.7: 13.42 (8.41) - Change in real education spending per capita (%): - All LICs: 0.85 (3.16) - LICs with weaker fundamentals: -1.85 (4.07) - PS>0.7: -3.45 (5.25) - Change in REER: - All LICs: -3.44 (5.8) - LICs with weaker fundamentals: -3.40 (6.59) - PS>0.7: -5.28 (7.66) - ODA commitments (% of GDP, level): - All LICs: 2.19*** (0.61) - LICs with weaker fundamentals: 1.78** (0.7) - PS>0.7: 1.92*** (0.75) - ODA disbursements (% of GDP, level): - All LICs: 1.55*** (0.52) - LICs with weaker fundamentals: 1.31** (0.57) - PS>0.7: 1.36** (0.62) - Changes in macroeconomic outcomes (X(t)-X(t-1)): - Change in real GDP growth (%): - All LICs: 0.74 (0.75) - LICs with weaker fundamentals: 1.41 (0.87) - PS>0.7: 2.21** (1) - Change in inflation (%): - All LICs: -6.45 (4.6) - LICs with weaker fundamentals: -10.47* (5.33) - PS>0.7: -13.96** (6.22) - Change in reserve coverage (months of imports): - All LICs: 0.71*** (0.1) - LICs with weaker fundamentals: 0.93*** (0.11) - PS>0.7: 0.79*** (0.12) - Change in current account balance plus FDI (% of GDP): - All LICs: 1.44** (0.71) - LICs with weaker fundamentals: 2.03** (0.84) - PS>0.7: 2.43** (0.97) - Change in government balance (% of GDP): - All LICs: 0.78* (0.46) - LICs with weaker fundamentals: 1.06* (0.55) - PS>0.7: 1.41** (0.59) - Change in ODA commitments (% of GDP): - All LICs: -0.24 (0.72) - LICs with weaker fundamentals: -0.98 (0.87) - PS>0.7: -1.04 (1.01) - Change in ODA disbursements (% of GDP): - All LICs: -0.15 (0.42) - LICs with weaker fundamentals: -0.57 (0.49) - PS>0.7: -0.73 (0.57) - Number of observations reported: 1633 / 349 / 293 (table shows "Number of observations /1633349293" corresponding to columns). - Significance notation: Significant at 10 percent:*; 5 percent:**; and 1 percent:***. Standard errors in parentheses. ### Interpretation and mechanisms - Short-term IMF-supported programs lead to significantly better outcomes especially for LICs with substantial prior macroeconomic imbalances and/or severe adverse external shocks. - Improvements include higher current account balances, higher reserve coverage, lower inflation, and lower fiscal deficits relative to control groups. - Program countries tend to have more depreciated real exchange rates, but differences with control groups are not significant. - Changes in real health and education spending per capita are not statistically different from control group changes. - Estimated positive impact on growth may be attributed to: - IMF financing easing short-term adjustment burdens. - Restoration of macroeconomic stability, especially for countries with significant prior instability. - Potential catalytic effects on Official Development Assistance (ODA), though empirical detection of catalytic role is limited in aggregate results. ### Findings on ODA and catalytic effects - Both commitments and disbursements of ODA are significantly higher for the program group. - Differences in disbursements are lower than differences in commitments compared to the control group, suggesting room to improve utilization and predictability of ODA for program countries. - No significant change in ODA detected as a definitive catalytic impact of IMF programs in some specifications; possible explanations: - Countries with high propensity scores may avoid or delay requesting IMF assistance because of ad hoc increases in ODA flows. - ODA provided as budget support may be more responsive to IMF programs than project support. - Conditioning matching on propensity score and ODA disbursements yields qualitatively similar levels results, and estimated impacts (including on growth) get stronger for program countries; however, differences in improvement in growth and government balances become insignificant after controlling for ODA disbursements. ### Robustness checks - Four sensitivity analyses: - (i) Relax implementation-record adjustment (interruptions of six months or longer). - (ii) Restrict sample to 1980–1999. - (iii) Condition matching on propensity score and ODA disbursements. - (iv) Condition matching on propensity score and lagged GDP growth. - Results are robust to these adjustments with some changes: - Not adjusting the IMF program dummy for implementation record yields qualitatively similar impacts but generally smaller magnitudes. - 1980–1999 sample: results qualitatively similar and quantitative impacts on growth and other indicators even stronger. - Conditioning on ODA disbursements: levels qualitatively similar; estimated impact stronger for program countries; changes in growth and government balances become insignificant, suggesting catalytic effect of IMF programs may be an important channel. - Conditioning on lagged GDP growth: participation equation shows lagged GDP growth strongly raises likelihood of a program request; conditioning on lagged growth yields qualitatively similar results and makes the impact on growth and improvement in growth significant for the whole sample and high propensity group. ### Sensitivity analysis to hidden bias (Rosenbaum) - Rosenbaum’s sensitivity analysis used to test robustness to hidden bias; parameter Gamma ( ) measures how much hidden bias can be present before conclusions change. - For countries with high propensity scores: - Results are less sensitive to hidden bias: odds ratios would need to increase in the range of 1.5-3.5 times before results become insignificant. - For the full sample: - Five variables (having less than 1.2) appear highly sensitive to hidden bias. - Reserve coverage and change in reserve coverage are the least sensitive. - Results for inflation are highly sensitive to hidden bias for both the full sample and the high propensity sample. - Note: Rosenbaum’s measure has limitations noted in literature (Robins (2002)) and its interpretation requires expert judgment about plausible ranges for . ### Conclusions and policy implications - Two distinct beneficial channels of IMF program support for LICs: - Longer-term policy support: - Associated with higher long-term growth rates, less growth volatility, more rapid reductions in poverty and inequality, higher government balances, higher levels of social spending, higher FDI, and lower inflation. - This result does not seem to depend on the amount of IMF financing provided over the longer term. - Short-term liquidity support: - Positively associated with higher short-term growth, current account balances, and reserve coverage, as well as lower inflation and fiscal deficits, especially pronounced for high propensity score countries with immediate balance of payments problems. - IMF financing combined with potential catalytic effects encouraging additional aid flows can provide a buffer to absorb shocks and prevent procyclical cuts in spending and investment. - Relevance to the global financial crisis: - Longer-term IMF support via successive medium-term programs (ECF and predecessors; PSI more recently) helped LICs raise longer-term growth and build macroeconomic buffers and institutions. - IMF’s sharp increase in financial assistance in 2009—doubling access and increasing commitments to roughly four times the historical average, plus the global SDR allocation—helped relax liquidity constraints and allowed LICs to preserve or increase spending during the crisis. - Combination of stronger pre-crisis buffers and crisis financing enabled most LICs to mount a countercyclical fiscal policy response in 2009, facilitating rapid economic recovery. - Policy inference: - IMF facilities for LICs should include a diverse set of tools: medium-term policy support (ECF, PSI) and quick short-term financing (augmentations of the ECF, SCF, RCF) when urgent balance of payments needs arise. - In absence of shocks, ECF arrangements and precautionary SCF arrangements can provide policy support and insurance even at low access levels. *Source: IMF staff calculations and empirical analysis presented in the supplied content.* ### Annex I. Propensity Score Matching (PSM) Methodology ### Annex I. Propensity Score Matching (PSM) Methodology ### A. Addressing Selection Bias—Alternative Approaches - Literature uses multiple approaches to construct a credible counterfactual for the impact of IMF-supported programs: - Before-after approach: assumes conditions affecting performance are the same before and after a program; attributed change is due to the program. Noted bias: changes in economic structure or unrelated shocks between periods (Ghosh and others, 2005). - Instrumental variables: instruments correlated with treatment selection but not the outcome (Barro and Lee, 2005). Major challenge: identifying appropriate and truly exogenous instruments. - Generalized Evaluation Estimator (GEE): uses policy reaction functions for non-program countries to approximate the counterfactual (Goldstein and Montiel, 1986). Criticism: Dicks-Mireaux, Mecagni, and Schadler (2000) largely discredit GEE due to many restrictive assumptions; report that the counterfactual policy reaction function does not have significant explanatory power for the sample of nonprogram observations. - Heckman selection correction model: two-stage approach (first stage probit for program engagement probability; second stage includes inverse Mills ratio). Requirement: an exclusion restriction—at least one explanatory variable influences selection but not the outcome. The inverse Mills ratio drops out only if correlation between unobserved determinants of participation and unobserved determinants of outcome is 0. - Propensity Score Matching (PSM): final method used in this paper (described below). ### B. PSM Methodology - Two-step statistical comparison: - Step 1 (selection model): Estimate the probability of participating in IMF-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 to construct a statistical control group. - Matching rationale: - Matching on likelihood of participation assures similarity of initial macroeconomic conditions and country characteristics between treated and control groups. - The control group proxies the counterfactual (macroeconomic outcomes if program countries had not had a program). - Program effects are calculated as the mean difference in macroeconomic outcomes across matched groups. - Key assumptions for identification: - Conditional independence (confoundedness): program participation is based entirely on observed pre-treatment characteristics. If unobserved characteristics determine participation, conditional independence is violated and PSM is inappropriate. - Common support: treated observations must have comparison observations “nearby” in the propensity score distribution. - Treatment definition in this study: - IMF engagement is treated as treatment status (analogous to microeconomic program evaluation literature). Treated = countries with IMF engagement; control = remaining countries. - Average treatment effect on the treated (ATT) is defined as the mean difference between observed outcomes and the counterfactual outcomes for treated units (notation in source: ATT = E[Yi1 − Yi0 | Di = 1], with Di the engagement dummy). - Matching algorithms used and robustness checks: - Nearest neighbor matching: primary technique; constructs control group by choosing three, four, and five nearest non-program countries by propensity score. - Radius matching: uses all comparison observations within a predefined distance around the propensity score. - Kernel matching: weighted average of outcomes of all nontreated units, weights related to proximity to treated unit. - Caution on interpretation: - PSM is useful when only observed pre-treatment characteristics affect program participation; using a rich set of pre-program data helps support conditional independence. - A well-specified and comprehensive selection model explaining IMF participation is key to properly assess program impact. ### C. Specification of the PSM Selection Model - General approach: - Focus on low-income countries (LICs); distinguish longer-term engagement from short-term financing to create more homogeneous samples for robust identification of participation determinants. - Selection model for longer-term IMF engagement: - Model: pooled probit regression. - Dependent variable: dummy = 1 if a country had five or more years of IMF-supported programs in a 10-year period; 0 otherwise. - Qualifying programs: IMF financial arrangements available to LICs (primarily ECF and predecessors PRGF, ESAF, SAF; also SBA, ESF-HAC, SCF; and nonfinancial PSI). Program years purged of episodes with prolonged interruptions. - Data treatment: analysis based on decadal averages where periods share a 50 percent overlap to increase observations. Decadal periods used: 1986–95; 1991–2000; 1996–2005; 2001–10. - Determinants included: initial macroeconomic buffers (reserve coverage and foreign aid to GDP ratio at beginning of decade), structural characteristics (geographic and institutional dummy variables), external demand (trading partners’ real GDP growth), and IMF quota (proxy for access to IMF resources). - Empirical findings (Table 5 coefficients and notes preserved): - Initial reserves -0.156*** (0.04) - Initial aid/GDP 0.016* (0.01) - Trading partner growth -0.092 (0.07) - IMF quota/GDP -0.078** (0.03) - Resource rents/GDP -0.020** (0.01) - Landlockedness 0.741*** (0.21) - Political connectedness 0.010* (0.01) - Polity 0.010 (0.02) - Constant 0.362 (0.52) - Observations 203 - Note: Robust standard errors in parentheses. Definition: a country is considered to have longer-term engagement in a given decade if in five or more years it had a financial arrangement or a Policy Support Instrument in place, for at least six months in each of these years. Significance levels: *10 percent; **5 percent; ***1 percent. - Interpretation of findings: - Higher initial reserves and lower aid associated with lower propensity for longer-term engagement. - Lower trading partner economic growth tends to increase propensity for longer-term engagement. - Landlocked and resource-poor countries have higher propensity for longer-term engagement. - Larger quota and lower political connectedness imply lower probability of longer-term engagement. - Selection model for short-term IMF engagement (demand for IMF financing in response to shocks): - Basis: draws on Bal Gündüz (2009); focuses on LIC arrangements addressing immediate balance of payments needs from policy and/or external shocks. - Dependent variable: panel dummy = 1 if a new IMF arrangement is approved (SBA, SAF/ESAF/PRGF/ECF augmentations, ESF, SCF, RCF, CFF), with refinements: - Precautionary SBA/SCF and SBA/PRGF/ECF augmentations addressing natural disasters are excluded. - Some SAF/ESAF/PRGF/ECF arrangements are added if they address immediate balance of payments needs arising from policy shocks (identified via program interruptions and IMF staff narratives). - Normal episodes defined as initial year of two successive years with no IMF financing for shocks when the member is eligible to access IMF resources; several refinements exclude episodes with arrears, natural disaster financing, program interruptions, SMPs, EPCAs, and other specified cases. - Estimation: binary response model for panel data (probit family); various estimators considered for panel heterogeneity (pooled probit, random effects probit, fixed effects probit); a correlated random effects probit model is preferred. - Key empirical results from Bal Gündüz (2009) (Table 6 coefficients and notes preserved): - Current account balance to GDP (t-1) -0.076*** (-4.61) - Reserve coverage in months of imports (CFA) (t-1) -0.478*** (-6.08) - Reserve coverage in months of imports (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 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: - Total debt service to exports 0.044*** (2.63) - FDI to GDP -0.105* (-1.76) - Pseudo R2 0.58 - Number of observations 532 - Number of countries 55 - Sample probability 0.44 - Note: Demand for IMF financing in response to policy and/or exogenous shocks (excluding natural disasters) is estimated by a correlated random effects probit model. Significant at 10 percent:*; 5 percent:**; and 1 percent:***, t-statistics in parenthesis. - Interpretation of findings: - Increased probability of IMF financing associated with lower reserve coverage, more negative current account balance, higher macroeconomic instability (mitot), lower real GDP growth, adverse terms of trade changes, and certain global shocks (change in real oil prices, cyclical world trade component). - Demand for IMF resources by LICs is likely cyclical in response to global conditions, intensity depending on magnitude and persistence of adverse external shocks. - Treatment variable and sample refinements for short-term analysis: - Treatment: panel dummy = 1 for approval of IMF-supported programs addressing immediate balance of payments needs; 0 for non-program episodes. - Refinements mirror those in selection equation; severe state failure events are excluded (identified via PITF dataset using SFTPMMAX > 3.9). - Years of program interruptions excluded to account for program implementation. - Some asymmetries introduced between treatment and participation dependent variable to increase common support (treatment includes nonprogram years followed immediately by a program and nonprogram episodes without IMF membership as zeros, whereas participation equation excluded these). - Macroeconomic stability indicator: - Composite index variant of Jaramillo and Sancak (2009) used (version including black market premium first used in Bal Gündüz (2009)). Formula and component definitions preserved in source text; higher mitot indicates increased macroeconomic instability. *Source: Annex I. Propensity Score Matching (PSM) Methodology, _wp13273 - Annex I. Propensity Score Matching (PSM) Methodology* ### Annex II. Panel Regression on the Determinants of Long-Term Growth ### Annex II. Panel Regression on the Determinants of Long-Term Growth ### Methodology and Estimation Strategy - Dynamic two-way fixed-effects models for panel data are used to compute the impact of IMF-supported programs on per capita GDP growth. - GMM estimators applied: Holtz-Eakin, Newey, and Rosen (1988); Arellano and Bond (1991); Arellano and Bover (1995). - System-GMM estimator is used (combines level and first-difference equations to increase moment conditions). - Internal instruments: previous observations of explanatory and lagged-dependent variables. - Diagnostic tests systematically applied: - Autocorrelation of second order of the residuals in differences. - Hansen over-identification test. - Limiting the number of internal instruments below the number of countries to avoid “instrument proliferation”. - All regression specifications control for the inverse Mills ratio to address selection bias. ### Model Specifications - Outcome variable: - refers to the real GDP per capita growth rate. - Baseline specification controls for Z, the matrix of control variables chosen not to be related to IMF engagement. - IMF longer-term dummy measures the total effect of IMF-supported programs on the level of growth. - Augmented specification includes potential transmission channel variables Y (those significantly affected by the IMF program dummy in the PSM estimations) to assess indirect channels: - If inclusion of Y lowers (in absolute terms) the magnitude and significance of the coefficient associated with the IMF dummy, Y is interpreted as a transmission channel. ### Definitions and Sample - Sample for descriptive figures: 75 low-income countries (LICs). - Longer-term engagement definition (figures and matching analyses): - 10 or more years of having an IMF financial arrangement or Policy Support Instrument in place during 1991–2010, for at least six months in each of these years (Figure 3 note). - Alternative decade-level definition used in some analyses: in five or more years a financial arrangement or a PSI in place, for at least six months in each of these years (Tables notes). - Period coverage in various analyses: - Decadal period averages between 1986 and 2005 (overlap by 50 percent). - Two 10-year period averages between 1996 and 2010 (overlap by 50 percent). - Four 10-year period averages between 1986 and 2010 (overlap by 50 percent). - Short-term engagement sample composed of 58 LICs and covers 1980–2010 (and subperiods 1980–1999). ### Diagnostic and Robustness Practices - Bootstrapped standard errors reported in matching estimations. - Significance notation used consistently: - * 10 percent significance; **5 percent significance; ***1 percent significance. - Rosenbaum sensitivity analysis reported with Γ values and associated probabilities (Table 14). ### Key Quantitative Results — Longer-Term IMF Engagement (selected coefficients) - Table 7 (Pre-2000s and Post-2000s; coefficients reported for four matching methods; bootstrapped standard errors in parentheses): - GDP per capita growth: 0.84 (Nearest-Neighbor), 1.19 (Five-Nearest-Neighbor), 1.60** (Radius), 1.42* (Kernel) — standard errors (0.96)(0.87)(0.72)(0.82). - GDP per capita growth volatility: -4.11*; -3.36*; -1.72**; -3.05* — std. errors (2.26)(1.80)(0.82)(1.83). - Government balance: 3.71*; 3.23; 3.87***; 4.20** — std. errors (2.22)(1.97)(1.39)(1.83). - Reserve coverage: 1.26**; 0.83; 0.84*; 0.67 — std. errors (0.60)(0.52)(0.44)(0.55). - Social spending and Education spending showed positive coefficients in some methods (e.g., Education spending 1.34; 1.91; 0.99*; 1.54). - Table 8 (Non-Overlapping Periods; bootstrapped standard errors in parentheses): - GDP per capita growth: 4.75***; 3.97***; 3.02**; 3.96*** — std. errors (1.26)(1.24)(1.17)(1.25). - Current account: -3.69*; -2.34; -1.83; -2.86 — std. errors (2.17)(1.93)(1.70)(1.83). - FDI: 2.12*; 2.20**; 1.67*; 2.38** — std. errors (1.12)(0.98)(0.90)(1.02). - Social spending: 1.34**; 0.88; 1.25***; 0.89 — std. errors (0.63)(0.58)(0.48)(0.58). - Education spending: 0.78*; 0.67*; 1.08***; 0.65 — std. errors (0.42)(0.40)(0.38)(0.40). - Poverty gap and Poverty rate reported negative coefficients in some specifications (e.g., Poverty gap -4.35; Poverty rate -7.31). - Table 9 (No Adjustment for Program Implementation; four matching methods): - GDP per capita growth: 2.73***; 2.45***; 2.01***; 2.53*** — std. errors (0.74)(0.71)(0.63)(0.69). - FDI: 1.58***; 1.36***; 0.74; 1.40*** — std. errors (0.60)(0.50)(0.51)(0.48). - Social spending: 0.83; 1.24**; 1.34***; 1.11** — std. errors (0.57)(0.60)(0.39)(0.54). - Education spending: 0.58; 0.97**; 1.16***; 0.83* — std. errors (0.51)(0.49)(0.33)(0.47). - Gini: -5.72*; -3.91; -2.92; -5.41 — std. errors (3.18)(3.09)(2.09)(3.66). ### Key Quantitative Results — Short-Term IMF Engagement (selected tables) - Table 10 (Impact of Short-Term IMF Engagement With No Adjustment for Implementation; All LICs and LICs with Weaker Fundamentals; sample composed of 58 LICs; covers 1980–2010): - Real GDP growth (%): 0.53; 0.87; 1.33** — std. errors (0.49)(0.56)(0.62). - Inflation (%): -13.35***; -15.70***; -17.42*** — std. errors (3.92)(4.65)(5.27). - Reserve coverage (months of imports): 0.58***; 0.68***; 0.53*** — std. errors (0.13)(0.13)(0.13). - Current account balance plus FDI (% of GDP): 2.14**; 2.79***; 3.58*** — std. errors (0.86)(0.99)(1.12). - Government balance (% of GDP): 1.64***; 1.90***; 2.24*** — std. errors (0.58)(0.69)(0.77). - ODA commitments (% of GDP): 1.88***; 1.55***; 1.66*** — std. errors (0.53)(0.6)(0.63). - Change in macroeconomic outcomes (first differences) show similar patterns for Reserve coverage (0.59***; 0.74***; 0.63***). - Table 11 (Impact of Short-Term IMF Engagement, 1980–99): - Real GDP growth (%): 1.31*; 1.70**; 2.15** — std. errors (0.71)(0.77)(0.88). - Inflation (%): -16.04***; -16.83**; -19.47** — std. errors (6.24)(7)(8.06). - Reserve coverage (months): 0.51***; 0.59***; 0.52*** — std. errors (0.17)(0.17)(0.18). - Current account balance plus FDI (% of GDP): 2.91**; 2.98**; 3.91*** — std. errors (1.21)(1.3)(1.47). - Government balance (% of GDP): 2.95***; 3.29***; 3.87*** — std. errors (0.82)(0.92)(1.05). - Table 12 (Matching on Propensity Score and ODA Disbursements; 1980–2010; sample of 58 LICs): - Real GDP growth (%) dgdpog: 1.75***; 2.19***; 2.70*** — std. errors (0.6)(0.67)(0.74). - Inflation (%) infl100og: -27.03***; -28.30***; -27.04** — std. errors (9.22)(9.88)(10.97). - Reserve coverage (months) impcovog: 0.51***; 0.80***; 0.66*** — std. errors (0.16)(0.17)(0.17). - Current account balance plus FDI cabfdiy: 2.39**; 2.95**; 3.56*** — std. errors (1.14)(1.2)(1.35). - Government balance gbaltogdp100og: 3.33***; 3.68***; 3.54*** — std. errors (0.99)(1.07)(1.17). - Table 13 (Matching on Propensity Score and Lagged GDP Growth): - Real GDP growth (%) dgdpog: 1.23**; 1.59**; 2.11*** — std. errors (0.57)(0.66)(0.73). - Inflation (%) infl100og: -22.10***; -26.09***; -24.56*** — std. errors (5.77)(7.16)(8.47). - Reserve coverage impcovog: 0.76***; 1.05***; 0.81*** — std. errors (0.14)(0.15)(0.16). - Current account balance plus FDI cabfdiy: 2.13**; 2.97***; 4.20*** — std. errors (1.02)(1.16)(1.32). - Change in macroeconomic outcomes (first differences) similarly show positive and significant effects for Real GDP Growth and Reserve coverage. ### Sensitivity to Hidden Selection Bias - Table 14 (Rosenbaum Sensitivity Analysis): - Reported Γ and Probability pairs for All LICs and LICs with Weaker Fundamentals (selected entries): - Real GDP growth (%) All LICs: Γ 1.38 Probability 0.044; LICs with Weaker Fundamentals: Γ 1.76 Probability 0.046. - Inflation (%) All LICs: Γ 1.14 Probability 0.045; LICs with Weaker Fundamentals: Γ 1.27 Probability 0.048. - Reserve coverage (months) All LICs: Γ 2.11 Probability 0.046; LICs with Weaker Fundamentals: Γ 2.57 Probability 0.049. - Government balance (% of GDP) All LICs: Γ 1.65 Probability 0.043; LICs with Weaker Fundamentals: Γ 1.76 Probability 0.045. - Note: Γ measures how much hidden bias can be present (how much Γ can deviate from 1 before results begin to change). ### Figures — Macroeconomic Conditions and Changes (descriptive) - Figure 3 and Figure 4 present decade averages and groupings for multiple macroeconomic indicators for 75 LICs (1981–2010 and subperiods), including: - Real GDP per Capita Growth (annual averages), Inflation (median of averages), Government Balance (percent of GDP), Current Account + FDI (percent of GDP), Reserves (months of imports), Real GDP per Capita Volatility (averages of standard deviations), Tax Revenue (percent of GDP), Capital Spending (percent of GDP), External Debt (percent of GDP), Aid (percent of GDP), Social Spending (percent of GDP), CPIA, FDI (percent of GDP), Exports (percent of GDP), Poverty, Education Spending (percent of GDP), Health Spending (percent of GDP). - Figure 5 shows distributions (median, 25th, 75th percentiles) of changes in decadal averages across countries by whether they have LT engagement (decadal definition: in five or more years had a financial arrangement or PSI in place for at least six months). - Figures 6 and 7 present estimated impacts of short-term Fund engagement relative to matched controls by propensity score thresholds (All; PS>0.5; PS>0.7) for levels and changes of macroeconomic outcomes. *Source: IMF staff calculations.* ### REFERENCES ### _wp13273 - REFERENCES ### Methodology and Econometrics - Aakvik, Arild, 2001, “Bounding a Matching Estimator: The Case of a Norwegian Training Program,” Oxford Bulletin of Economics and Statistics, Vol. 63 (1), 115–43. - Arellano, Manuel, and Stephen Bond, 1991, “Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations,” Review of Economic Studies, 58 (2), pp. 277–97. - Arellano, Manuel, and Olympia Bover, 1995, “Another Look at the Instrumental Variable Estimation of Error-Components Models,” Journal of Econometrics, 68 (1), pp. 29–51. - Chamberlain, Gary, 1982, “Multivariate Regression Models for Panel Data,” Journal of Econometrics, Vol. 18 (1), pp. 5–46. - Holtz-Eakin, Douglas, Whitney Newey, and Harvey S. Rosen, 1988, “Estimating Vector Autoregressions with Panel Data,” Econometrica, 56 (6), pp. 1371–95. - Mundlak, Yair, 1978, “On the Pooling of Time Series and Cross Section Data,” Econometrica, Vol. 46 (1), pp. 69–85. - Robins, James M., 2002, Comment on “Covariance Adjustment in Randomized Experiments and Observational Studies,” Statistical Science, Vol. 17 (3), pp. 309–21. - Rosenbaum, Paul R., 2002, “Observational Studies,” Springer Series in Statistics (New York). - Rosenbaum, Paul R., and Donald B. Rubin, 1983, “The Central Role of the Propensity Score in Observational Studies for Causal Effects,” Biometrika, Vol. 70 (1), pp. 41–55. ### IMF Program Evaluation, Impact, and Outcomes - Atoyan, Ruben, and Patrick Conway, 2006, “Evaluating the Impact of IMF Programs: A Comparison of Matching and Instrumental-Variable Estimators,” The Review of International Organizations, Vol. 1 (2), pp. 99–124. - Bal Gündüz, Yasemin, 2009, “Estimating Demand for IMF Financing by Low-Income Countries in Response to Shocks,” IMF Working Paper No. 09/263 (Washington: International Monetary Fund). - Bal Gündüz, Yasemin, Christian Ebeke, Burcu Hacibedel, Linda Kaltani, Vera Kehayova, Chris Lane, Christian Mumssen, Nkunde Mwase, and Joseph Thornton, “The Economic Impact of IMF-Supported Programs in Low-Income Countries,” IMF Occasional Paper No. 13/277 (Washington: International Monetary Fund). - Barro, Robert and Jong-Wha Lee, 2005, “IMF Programs: Who is Chosen and What Are the Effects?” Journal of Monetary Economics, Vol. 52 (7), pp. 1245–69. - Bordo, Michael D., and Anna J. Schwartz, 2000, “Measuring Real Economic Effects of Bailouts: Historical Perspectives on How Countries in Financial Distress have Fared With and Without Bailouts,” Carnegie-Rochester Conference Series on Public Policy, Vol. 53, No. 1, pp. 81–167. - Butkiewicz, James L., and Halit Yanikkaya, 2005, “The Effects of IMF and World Bank Lending on Long-Run Economic Growth: An Empirical Analysis,” World Development, Vol. 33 (3), pp.371–91. - Clements, Benedict, Sanjeev Gupta, and Masahiro Nozaki, 2011, “What Happens to Social Spending in IMF-Supported Program?,” IMF Staff Discussion Note No. 11/15 (Washington: International Monetary Fund). - ———, 2012, “What Happens to Social Spending in IMF-Supported Programs?” Applied Economics, Vol. 45 (28), pp. 4022–33. - Conway, Patrick, 1994, “IMF Lending Programs: Participation and Impact,” Journal of Development Economics, Vol. 45 (2), pp. 365–91. - Dicks-Mireaux, Louis, Mauro Mecagni, and Susan Schadler, 2000, “Evaluating the Effect of IMF Lending to Low-Income Countries,” Journal of Development Economics, Vol. 61, pp. 495–526. - Dreher, Axel, 2006, “IMF and Economic Growth: The Effects of Programs, Loans, and Compliance with Conditionality,” World Development, Vol. 34 (5), pp. 769–88. - Evrensel, Ayse, 2002, “Effectiveness of IMF-Supported Stabilization Programs in Developing Countries,” Journal of International Money and Finance, Vol. 21 (5), pp. 565–87. - Hajro, Zlata, and Joseph P. Joyce, 2009, “A True Test: Do IMF Programs Hurt the Poor?,” Applied Economics, 41 (3), pp. 295–306. - Hardoy, Inés, 2003, “Effect of IMF Programmes on Growth: A Reappraisal Using the Method of Matching,” Institute for Social Research Paper, 2003:040. - Hutchison, Michael M., 2003, “A Cure Worse that the Disease? Currency Crises and the Output Costs of IMF-Supported Stabilization Programs,” In M.P. Dooley & F. A. Jeffrey (eds.), Managing Currency Crises in Emerging Markets, pp. 321–59 (Chicago: University of Chicago Press). - ———, 2004, “Selection Bias and the Output Costs of IMF Programs,” EPRU Working Paper Series 04–15, Economic Policy Research Unit (EPRU), Department of Economics (Denmark: University of Copenhagen). - Oberdabernig, Doris A., 2013, “Revisiting the Effects of IMF Programs on Poverty and Inequality,” World Development, 46, pp. 113-42. - Przeworski, Adam and James Raymond Vreeland, 2000, “The Effect of IMF Programs on Economic Growth,“ Journal of Development Economics, Vol. 62 (2), pp. 385–421. - Ul Haque, Nadeem, and Mohsin S. Khan, 1998, “Do IMF-Supported Programs Work? A Survey of the Cross-Country Empirical Evidence,” IMF Working Paper No. 98/169 (Washington: International Monetary Fund). ### Political Economy, Conditionality, and Determinants of Lending - Andersen, Thomas Barnebeck, Henrik Hansen, and Thomas Markussen, 2006, ”US Politics and World Bank IDA-Lending,” Journal of Development Studies, Vol. 42, No. 5, pp. 772–94. - Bird, Graham, 2007, “The IMF: A Bird’s Eye View of its Role and Operations,” Journal of Economic Surveys, Vol. 21 (4), pp 683–745. - ———, and Paul Mosley, 2006, “Should the IMF Discontinue Its Long-term Lending Role in Developing Countries?” In Globalization and the Nation State: the Impact of the IMF and the World Bank, edited by Gustav Ranis, James Raymond Vreeland, and Stephen Kosack. London: Routledge. - ———, Mumtaz Hussain, and Joseph P. Joyce, 2004, “Many Happy Returns? Recidivism and the IMF,” Journal of International Money and Finance, Vol. 23 (2), pp. 231–51. - ———, and Dane. Rowlands, 2001, “IMF Lending: How Is It Affected by Economic, Political and Institutional Factors?” The Journal of Policy Reform, Vol. 4 (3), pp. 243–70. - ———, 2007, “The IMF and the Mobilisation of Foreign Aid,” Journal of Development Studies, Vol. 43, No. 5, pp. 856–70, July 2007. - ———, 2009, “A Disaggregated Empirical Analysis of the Determinants of IMF Arrangements: Does One Model Fit All?” Journal of International Development, Vol. 21 (7), pp. 915–31. - Dreher, Axel, and Nathan M. Jensen, 2007, “Independent Actor or Agent? An Empirical Analysis of the Impact of US Interests on IMF Conditions,” Journal of Law & Economics, Vol. 50 (1), pp. 105–24. - Dreher, Jan-Egbert Sturm, and James Raymond Vreeland, 2006, “Does Membership on the UN Security Council Influence IMF Decisions? Evidence from Panel Data,” CESifo Working Paper Series No. 1808, CESifo Group Munich. - Dreher, and Roland Vaubel, 2004, “Do IMF and IBRD Cause Moral Hazard and Political Business Cycles? Evidence from Panel Data,” Open Economies Review, Vol. 15 (1), pp. 5–22. - Garuda, Gopal, 2000, “The Distributional Effects of IMF Programs: A Cross-Country Analysis,” World Development, 28 (6). 1031–51. - Ghosh, Atish, Charalambos Christofides, Jun I. Kim, Laura Papi, Uma Ramakrishnan, Alun Thomas, and Juan Zalduendo, 2005, The Design of IMF-Supported Programs, IMF Occasional Paper No. 241 (Washington: International Monetary Fund). - Joyce, Joseph P., 1992, “The Economic Characteristics of IMF Program Countries,” Economic Letters, Vol. 38 (2), pp. 237–42. - ———, 2005, “Time Past and Time Present: A Duration Analysis of IMF Program Spells,” Review of International Economics, Vol. 13 (2), pp. 283–97. - Mecagni, Mauro, 1999, “The Causes of Program Interruptions,” In Hugh Bredenkamp and Susan Schadler, eds.1999, Economic Adjustment and Reform in Low-Income Countries, Studies by the Staff of the International Monetary Fund, Chapter 9, pp 215–276 (Washington: International Monetary Fund). - Moser, Christoph, and Jan-Egbert Sturm, 2011, "Explaining IMF Lending Decisions after the Cold War," Review of International Organizations, Vol. 6 (3), pp. 307–40. - Nooruddin, Irfan, and Joel W. Simmons, 2006, “The Politics of Hard Choices: IMF Programs and Government Spending,” International Organization, Vol. 60, No. 4, pp. 1001–33. - Oatley, Thomas, and Jason Yackee, 2004, “American Interests and IMF Lending,” International Politics, Vol. 41, pp. 415–29. - Presbitero, Andrea F., and Alberto Zazzaro, 2012, “IMF Lending in Times of Crisis: Political Influences and Crisis Prevention,” World Development, Vol. 40, No. 10, pp. 1944–69. - Santaella, Julio A., 1996, “Stylized Facts Before IMF-Supported Macroeconomic Adjustment,” IMF Staff Papers No. 43, pp. 502–44 (Washington: International Monetary Fund). - Steinwand, Martin, and Randall Stone, 2008, “The International Monetary Fund: A Review of the Recent Evidence,” Review of International Organizations, Vol. 3 (2), pp. 123–49. - Stone, Randall W., 2002, Lending Credibility: The International Monetary Fund and the Post-Communist Transition (Princeton: Princeton University Press). - ———, 2004, “The Political Economy of IMF Lending in Africa,” American Political Science Review, Vol. 98 (4), .pp. 577–91. - Sturm, Jan-Egbert, Helge Berger, Jakob de Haan, 2005, “Which Variables Explain Decisions on IMF Credit? An Extreme Bounds Analysis,” Economics & Politics, Vol. 17 (7), pp. 177–213. - Vreeland, James Raymond, 2003, “The IMF and Economic Development” (Cambridge: Cambridge University Press). ### Low-Income Countries, Facilities, and Policy Reviews - Bredenkamp, Hugh and Susan Schadler, eds., 1999, “Economic Adjustment and Reform in Low-Income Countries: Studies by the Staff of the International Monetary Fund” (Washington: International Monetary Fund). - Cerutti, Eugenio, 2007, “IMF Drawing Programs: Participation Determinants and Forecasting,” IMF Working Paper No. 07/152 (Washington: International Monetary Fund). - Fabrizio, Stefania, 2009, “Coping with the Global Financial Crisis: Challenges Facing Low-Income Countries” (Washington: International Monetary Fund). - Ghosh, Atish, et al., 2005, The Design of IMF-Supported Programs, IMF Occasional Paper No. 241 (Washington: International Monetary Fund). - International Monetary Fund, 2006, Review of Ex Post Assessments and Issues Relating to the Policy on Longer-Term Program Engagement (Washington). - ———, 2009, The Fund’s Facilities and Financing Framework for Low-Income Countries—Supplementary Information (Washington). - ———, 2010, Emerging from the Global Crisis: Macroeconomic Challenges Facing Low-Income Countries (Washington). - ———, 2012a, “Review of Facilities for Low-Income Countries,” (Washington: International Monetary Fund). - ———, 2012b, “Review of Facilities for Low-Income Countries—Supplement 1” (Washington: International Monetary Fund). - ———, 2012c, “2011 Review of Conditionality—Overview Paper” (Washington: International Monetary Fund). - ———, 2012d, “2011 Review of Conditionality—Background Paper: Outcomes of IMF-Supported Programs” (Washington: International Monetary Fund). - ———, 2013, “Review of Facilities for Low-Income Countries—Proposals for Implementation” (Washington: International Monetary Fund). - Independent Evaluation Office, 2002, Evaluation of Prolonged Use of IMF Resources (Washington: International Monetary Fund). - ———, 2004, Evaluation of the IMF’s Role in Poverty Reduction Strategy Papers and the Poverty Reduction and Growth Facility (Washington: International Monetary Fund). - Ivanova, Anna, Wolfgang Mayer, Alex. Mourmouras, and George Anayiotos, 2003, “What Determines the Implementation of IMF-Supported Programs?” IMF Working Paper No. 03/8, Published in A. Mody and A. Rebucci (eds), IMF-Supported Programs: Recent Staff Research 2006 (Washington: International Monetary Fund). - Jaramillo, Laura and Cemile Sancak, 2009, Why Has the Grass Been Greener on One Side of Hispaniola? A Comparative Growth Analysis of the Dominican Republic and Haiti, IMF Staff Papers, Vol. 56, pp. 323–49 (Washington: International Monetary Fund). - Mecagni, Mauro, 1999, “The Causes of Program Interruptions,” In Hugh Bredenkamp and Susan Schadler, eds.1999, Economic Adjustment and Reform in Low-Income Countries, Studies by the Staff of the International Monetary Fund, Chapter 9, pp 215–276 (Washington: International Monetary Fund). - Mercer-Blackman, Valerie, and Anna Unigovskaya, 2000, “Compliance with IMF Program Indicators and Growth in Transition Economies,” IMF Working Paper No, 00/47 (Washington: International Monetary Fund). ### Sectoral and Country Studies; Complementary Analyses - Chari, Anusha, Wenjie Chen, and Kathryn Dominguez, 2012, “Foreign Ownership and Firm Performance: Emerging Market Acquisitions in the United States,” IMF Economic Review, Vol. 60 (1), pp. 1–42. - Clément, Matthieu, 2011, “Remittances and Household Expenditure Patterns in Tajikistan: A Propensity Score Matching Analysis, Asian Development Review, Asian Development Bank, Vol. 28 (2), pp. 58–87. - Garuda, Gopal, 2000, “The Distributional Effects of IMF Programs: A Cross-Country Analysis,” World Development, 28 (6). 1031–51. - Jaramillo, Laura and Cemile Sancak, 2009, Why Has the Grass Been Greener on One Side of Hispaniola? A Comparative Growth Analysis of the Dominican Republic and Haiti, IMF Staff Papers, Vol. 56, pp. 323–49 (Washington: International Monetary Fund). - Knight, Malcolm, and Julio A. Santaella, 1997, “Economic Determinants of IMF Financial Arrangements,” Journal of Development Economics, Vol. 54 (2), pp. 405–36. - Lin, Shu, 2010, “On the International Effects of Inflation Targeting,“ The Review of Economics and Statistics, Vol. 92 (1), pp. 195–99. - ———, and Haichun Ye, 2007, “Does Inflation Targeting Really Make a Difference? Evaluating the Treatment Effect of Inflation Targeting in Seven Industrial Countries,“ Journal of Monetary Economics, Vol. 54 (8), pp. 2521–33. - ———, 2009, “Does Inflation Targeting Make a Difference in Developing Countries?“ Journal of Development Economics, Vol. 89 (1), pp. 118–23. - Marchesi, Silvia, and Emanuela Sirtori, 2011, “Is Two Better Than One? The Effects of IMF and World Bank Interaction on Growth,” The Review of International Organizations, Vol. 6, No. 3, pp. 287–306. - Moser, Christoph, and Jan-Egbert Sturm, 2011, "Explaining IMF Lending Decisions after the Cold War," Review of International Organizations, Vol. 6 (3), pp. 307–40. - Tapsoba, René, 2012, “Do National Numerical Fiscal Rules Really Shape Fiscal Behaviours in Developing Countries? A Treatment Effect Evaluation,” Economic Modelling, Vol. 29 (4), pp. 1356–69. *References as listed in _wp13273 - REFERENCES* --- _Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13273.pdf_