## _wp12232

## Source details

**Canonical URL:** [_wp12232](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12232.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12232.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12232.pdf.json)

---

### I. Introduction and research scope
- Structural reforms are essential to unlock economies’ growth potential.
- IMF support channels examined: surveillance, IMF programs (financial assistance plus policy prescriptions), and capacity building (training and technical assistance).
- Paper broadens scope beyond IMF programs’ economic impact to study IMF capacity building and its relation to IMF programs.
- Contributions:
  - Focus on capacity building, largely ignored in prior literature.
  - Investigates linkages between IMF capacity building and IMF programs; first study to focus on this relationship to authors' knowledge.
  - Uses structural reforms (sectoral regulatory indices) as dependent variables rather than distant outcomes such as growth and inflation.
  - Introduces a novel instrument for capacity building based on donor financing data to help isolate a causal relationship.

### II. Data
- Structural reforms dataset:
  - New dataset compiled by the Research Department of the IMF for a sample of 150 countries.
  - Indices cover six sectors: product and agriculture markets, trade, current account, domestic financial markets, capital account (indices normalized between 0–1; 1 = ‘most liberalized’).
- IMF training dataset:
  - IMF training data available from 1981 to 2011 (source: IMF Institute for Capacity Development’s Participant and Applicant Tracking System (PATS)).
  - Regression analysis uses number of officials trained; number of applicants used as a control for government willingness to build capacity.
  - Observed patterns:
    - Number of officials trained: flat in the 1980s; increased sharply during the 1990s for Central and Eastern Europe; steady increases for East Asia and Pacific and Sub-Saharan Africa in the mid 1990s.
    - Central, Eastern and South Asia received the most IMF training; Latin America and Sub-Sahara also received a relatively significant number.
- IMF technical assistance (TA) dataset:
  - TA data available for 1990–2012.
  - TA covers macroeconomic policy, tax policy and revenue administration, exchange rate system, financial sector stability, legislative frameworks, and macroeconomic and financial statistics.
  - Donor financing increasingly important for TA; donor financing information used to instrument TA.
  - Observed patterns:
    - Amount of IMF TA stable across regions except Central and Eastern Europe.
    - Sub-Saharan Africa, Central and Eastern Europe, and East Asia have received the highest amount of IMF TA.
- IMF programs:
  - IMF program data available for 1952-2012.
  - Evolution of IMF programs by region described for 1990-2012; evolution "pretty flat" over 1990-2012 except for Central and Eastern Europe.
  - Over the period, Sub-Saharan Africa, followed by Central and Eastern Europe and then Latin America, has the most important amount of IMF programs.
  - Appendix 1: typology of IMF programs distinguishes non-concessional and concessional lending facilities.

### III. Methodology and empirical specification
- Main estimation approaches:
  - Panel fixed effects estimation.
  - Difference- and system-GMM (Blundell and Bond, 1998).
  - Instrumental-variable strategies for TA and programs.
- Main regressors in structural reform model:
  - Prog = number of years under an IMF program.
  - Train = indicator whether share of officials trained is above/below median (training normalized by number of public sector officials; alternative thirds also used).
  - TA = number of person-years of technical assistance (contemporaneous TA used; training lagged).
  - Interaction terms: Prog × Train and Prog × TA.
- Program failure model:
  - Dependent variable PF: PF = 1 for whole duration if at least one year program was qualified as off track; PF = 0.5 if a canceled program was replaced immediately; PF = 0 if program was on track.
  - Estimation: OLS (and Logit as robustness) using yearly observations.
- Instrumental variables and controls:
  - Instrument for TA: donor contributions weighted by donors' bilateral aid "country preferences" (donor data from IMF Institute for Capacity Development; bilateral aid data 1981-2010 from OECD, 2012).
  - Instruments for programs: (i) UN voting alignment with G7/US; (ii) share of IMF staff from a certain nationality; (iii) a country's IMF quota.
  - Controls include: number of applications to IMF courses (motivation), Polity2 score (range -10 to +10), bureaucratic quality (range 0 to 4), average years of schooling (Barro and Lee, 2010), lagged level of regulation index.

### IV. Main empirical findings
- General associations:
  - Panel fixed effects: IMF programs associated with a significant increase in structural reforms in all sectors but trade and capital account in basic estimates (Table 1). These raw results lose robustness once controls are included.
- Training results:
  - IMF training leads to an increase in structural reforms only through IMF programs and only when a significant share of public servants is trained.
  - Table 2 (OLS with interactions):
    - Prog × Train statistically significant at the 5% level for:
      - Domestic financial sector reforms (column (1)),
      - Capital account (column (6)),
      - Trade reforms (column (3)).
    - Quantitative examples:
      - In a country with a 3 year IMF program, receiving above-median training (relative to below-median) implies an increase in domestic financial sector reforms by 0.099.
      - Trade reform would increase by 0.057.
      - Capital account reform would increase by 0.178.
    - Median IMF training in a 5 year window corresponds roughly to 200 officials trained (example context given).
  - Difference-GMM (Table 3): point estimates similar to Table 2; interaction Prog × Train even more statistically significant when excluding TA (increases sample by about 10 years).
  - System-GMM controlling for country motivation (Table 5): interaction Prog × Train remains positive and in several sectors significant (examples: Domestic Finance 0.0207* (0.0108); Networks 0.0590** (0.0230); Capital 0.0401* (0.0239)).
- Technical assistance results:
  - TA does not significantly lead directly to more structural reforms in main specifications.
  - TA is robustly associated with lower probability of program cancellation (Table 4: TA in last 5 years associated with fewer cancelled programs; examples of coefficients: -0.0142*** (0.00338); -0.0208*** (0.00566); -0.0233*** (0.00627)).
  - Instrumenting TA using donor-funding-based instrument (Table 6 and Table 7) confirms:
    - TA helps keep programs on track; IV estimates show negative association between TA and cancelled-program ratio (examples: -0.0374* (0.0205); -0.0390* (0.0230); -0.0382** (0.0191)), though precision varies across specifications.
  - Table 6 (system-GMM with donor-funding instrument): TA person-years coefficient examples include Domestic Finance: -0.0109** (0.00517); interaction TA×Program small but positive in some sectors (Domestic Finance: 0.00237** (0.00119); Networks: 0.00199** (0.00161)).
- Complementarity and interpretation:
  - Evidence indicates complementarity between IMF capacity building and IMF programs:
    - Training effective at raising structural reforms only when combined with IMF programs and when training reaches a critical mass of officials.
    - Technical assistance supports program implementation by reducing program failure risk rather than directly raising structural reform indices.

### V. Determinants and exploratory regressions (Appendix 2: Table A1)
- Determinants (selected significant associations, OLS with country-clustered SEs):
  - Number of program years → TA: 0.476*** (0.155).
  - Crisis Count → TA: 0.202*** (0.0640).
  - Crisis Count → Programs: 0.0885*** (0.0244).
  - Real GDP per capita → TA: -0.000104** (4.71e-05).
  - Real GDP per capita → Programs: -6.58e-05*** (1.68e-05).
  - Bureaucratic Quality → Training per gov. empl.: -0.0332* (0.0194).
  - Bureaucratic Quality → Programs: -0.346** (0.163).
  - Polity2 Score → Programs: 0.0671** (0.0311).
- Program association with structural reforms (Table 1, FE estimates):
  - Number of years with an active program positively associated with reforms in:
    - Domestic Finance: 0.0215*** (0.00387).
    - Current Account: 0.0127*** (0.00344).
    - Agriculture: 0.0245*** (0.00330).
    - Networks: 0.00770*** (0.00225).

### VI. Robustness, endogeneity, and identification
- Potential biases addressed:
  - Upward bias from unobserved government willingness to reform; proxied by number of applications to IMF courses.
  - Endogeneity between programs, TA, and reforms.
  - Downward bias if strong reformers do not use IMF capacity building.
- Identification strategies:
  - Country and year fixed effects, system GMM, instrumenting TA (donor-weighted contributions), and instruments for programs (UN voting alignment with G7/US; share of IMF staff from nationality; IMF quota).
  - Including lagged number of applications to IMF courses as a control for motivation (coefficient on applications positive across regressions).
- Robustness checks:
  - Results robust to:
    - Using an indicator equal to 1 if country had at least one IMF program over last 5 years.
    - Splitting training into three groups rather than two.
    - Splitting world into Advanced Economies and Central and Eastern Europe versus the rest.
    - Excluding crisis occurrences.
    - Restricting analysis to concessional programs (results remain robust; concessional programs have stronger reform focus).

### VII. Policy implications and recommendations
- Consolidate capacity building strategy and integrate it with IMF lending activities to strengthen reform windows of opportunity where program conditionality and government ownership align.
- Target training at potential program countries to yield the most impact.
- Prioritize delivering training en masse to reach a critical mass of officials; evidence suggests this fosters consensus and raises the likelihood of reforms in program contexts.
- Use technical assistance strategically to support implementation and reduce program failure risk.

_Italic: Source: _wp12232 - References; Appendix 1; Appendix 2 (extracted content provided)._

### References .............................................................................................................

### _wp12232 - References

### I. Introduction
- Structural reforms are essential to unlock economies’ growth potential.
- The International Monetary Fund (IMF) supports members through surveillance, IMF programs (financial assistance plus policy prescriptions), and capacity building (training and technical assistance).
- IMF Managing Director, Christine Lagarde, recently stressed that “Technical assistance and training are a core mandate of the Fund, along with surveillance and lending”.
- The paper broadens the scope beyond the economic impact of IMF programs to study the impact of IMF capacity building and its relation to IMF programs.
- Contributions of the paper:
  - Focuses on capacity building, which has been largely ignored in the literature.
  - Investigates linkages between IMF capacity building and IMF programs; first study to focus on this relationship to the authors' knowledge.
  - Uses structural reforms rather than distant outcome variables like growth and inflation as dependent variables.
  - Introduces a novel instrument for capacity building based on donor financing data to help isolate a causal relationship.

### Key research questions and scope
- Investigates the role that IMF programs and capacity building play in fostering structural reforms.
- Utilizes two novel datasets on IMF capacity building and structural reforms available for over one hundred IMF member countries over the period 1980 – 2010.
- Empirical strategy: panel fixed effects estimation and various robustness and endogeneity checks (detailed in Sections III and IV in the source).

### Main empirical findings (panel fixed effects estimation)
- First, there is a general association between IMF programs and structural reforms but this relationship is not very robust.
- Second, IMF training leads to an increase in structural reforms but only through IMF programs and only when a significant share of public servants is trained.
- Third, IMF technical assistance does not significantly lead to more structural reforms but raises the likelihood of completion of ongoing IMF programs.
- Overall conclusion: the paper provides robust empirical evidence of complementarity between IMF programs and IMF capacity building; capacity building activities increase the likelihood that a subsequent IMF program provides a window of opportunity for reforms in which IMF program conditionality and governments’ reform ownership are reconciled.

### II. Data
- Structural reforms dataset:
  - New dataset compiled by the Research Department of the IMF for a sample of 150 countries.
  - Indices describe the degree of regulation of six sectors covering both real sectors (product and agriculture markets, trade, and current account) and financial ones (domestic financial markets and capital account).
  - Each index contains different sub-indices summarizing dimensions of the regulatory environment in each sector.
  - Indices are normalized between 0–1, where 1 refers to ‘most liberalized’.
  - Ostry, Prati and Spilimbergo (2009) provide detailed discussion on index construction (as cited in source).

- IMF capacity building dataset:
  - Constructed an exhaustive and novel dataset of capacity building comprising both training activities and technical assistance.
  - IMF training data available from 1981 to 2011.
  - Training data source: IMF Institute for Capacity Development’s Participant and Applicant Tracking System (PATS) which tracks civil servants who participate in IMF Institute training courses. PATS provides data on country of residence, agency, age, gender, position, and detailed educational background of applicants.
  - Regression analysis uses number of officials trained from a given country to capture impact of training and number of applicants to control for government willingness to build capacity.
  - Training courses cover macroeconomic policy, national statistics and finance; offered regionally or at IMF headquarters in Washington, DC.
  - Observed patterns:
    - Number of officials trained: rather flat in the 1980s; increased sharply during the 1990s for Central and Eastern Europe; steady increases for East Asia and Pacific and Sub-Saharan Africa in the mid 1990s.
    - Central, Eastern and South Asia have received the most IMF training; Latin America and Sub-Sahara have received a relatively significant number too.

- IMF technical assistance dataset:
  - IMF technical assistance data available for the period 1990–2012.
  - Technical assistance covers macroeconomic policy, tax policy and revenue administration, exchange rate system, financial sector stability, legislative frameworks, and macroeconomic and financial statistics.
  - Donors play an increasingly important role in financing technical assistance; donor financing information is used to instrument technical assistance in regressions.
  - Observed patterns:
    - Amount of IMF technical assistance has been rather stable for all regions except Central and Eastern Europe.
    - Sub-Saharan Africa followed by Central and Eastern Europe and East Asia have received the highest amount of IMF technical assistance.
    - Central and Eastern Europe as well as Central and Eastern Asia and Sub-Saharan Africa are identified as the largest recipients of IMF technical assistance.

- IMF programs:
  - IMF programs are associated with financial lending arrangements; “A policy program supported by IMF financing is designed by the national authorities in close cooperation with the IMF, and continued financial support is conditioned on effective implementation of this program.”
  - Several IMF lending instruments exist and have been subject to a recent overhaul to make them more flexible (as stated in source).

### Methodology notes
- Uses panel fixed effects estimation.
- Introduces a novel instrument for capacity building based on donor financing data to address endogeneity for technical assistance.
- Regression specifications include controls for country motivation (e.g., applicants to training) and consider interactions between training/TA and IMF programs.
- Robustness: results are robust to a large number of checks, estimators and correcting for endogeneity (as stated in source).

### Implications and interpretation
- Evidence suggests complementarity between IMF capacity building and IMF programs:
  - Training appears effective at raising structural reforms only when combined with IMF programs and when training reaches a significant share of public servants.
  - Technical assistance appears to support program implementation by raising the likelihood of completion of ongoing IMF programs rather than directly increasing structural reforms.
- Policy relevance: consolidating capacity building strategy and integrating it with IMF lending activities may strengthen the reforms window of opportunity where program conditionality and government ownership align.

_Italic: Source: _wp12232 - References (extracted content provided)._

### Appendix 1 provides a list of existing IMF lending facilities.  IMF program data is available

### _wp12232 - Appendix 1 provides a list of existing IMF lending facilities. IMF program data is available

### Data coverage and regional patterns
- IMF program data available for the period 1952-2012.
- Evolution of IMF programs by regions is described for 1990-2012; evolution is "pretty flat" over 1990-2012 except for Central and Eastern Europe.
- Over the period, Sub-Saharan Africa, followed by Central and Eastern Europe and then Latin America, has the most important amount of IMF programs.
- IMF programs appear linked to IMF technical assistance; this pattern contrasts with training and programs where a similar pattern is not observed.

### Controls used in regressions
- Political institutions: revised combined Polity score (Polity2), range -10 to +10 (higher values indicate stronger political institutions).
- Bureaucratic quality: indicator from The International Country Risk Guide (2012), range 0 to 4 (higher value = stronger bureaucracy).
- Educational attainment: average years of schooling (Barro and Lee, 2010).
- Other controls: lagged level of the regulation index for associated sector (used to control convergence; estimates not reported).

### Determinants of IMF capacity building and programs (exploratory findings)
- Countries experiencing a crisis are associated with more IMF technical assistance and IMF programs, but not necessarily more IMF training.
- Member countries with higher bureaucratic quality are associated with less IMF training and less IMF programs.
- Higher GDP per capita is associated with less technical assistance and programs.
- Better political institutions (polity score) and higher foreign debt over GDP are associated with more programs.
- In general, more IMF programs are associated with more technical assistance but not more training.

### Conceptual channels: how capacity building may affect reforms
- IMF training channels:
  - Raises awareness of newest academic/policy developments and international best practices.
  - Familiarizes officials with IMF macroeconomic framework used in Article IV consultations.
  - When delivered to a large enough group of officials, can create consensus that reforms are needed, especially in the context of an IMF program.
  - Expectation: impact of IMF training depends on whether a significant share of officials is trained and on whether there is an ongoing IMF program.
- IMF technical assistance (TA) channels:
  - Less likely to influence decision to adopt reforms; more a supporting tool for implementation of ongoing reforms.
  - Narrow focus; improves implementation when political will to reform already exists.
  - Historically closely associated with IMF program conditionality; may increase likelihood of IMF program implementation and help avoid derailment.

### Empirical model specification (overview)
- Structural reforms model includes:
  - Country fixed effects δi and year fixed effects γt.
  - Main regressors: Prog (number of years under an IMF program), Train (indicator whether share of officials trained is above/below median), TA (number of person-years of technical assistance).
  - Interaction terms: Prog × Train and Prog × TA.
  - Training normalized by number of public sector officials (ILO, 2012); Train = 1 if fraction of officials trained is above median, 0 otherwise. Also alternative definitions: bottom/middle/upper thirds.
  - Use lagged values of Training; contemporaneous TA.
  - Estimators: OLS, difference- and system-GMM (Blundell and Bond, 1998).
  - Use 5-year averages to reduce noise and capture lags between reform decision and implementation.
- Program failure model:
  - Dependent variable PF: takes value 1 for whole duration if at least one year program was qualified as off track; 0.5 if a canceled program was replaced immediately; 0 if program was on track.
  - Estimate average marginal effect of Train and TA on PF using OLS (and Logit as robustness).
  - Use yearly observations for PF regressions.

### Key empirical results
- Table 1 (basic estimates): IMF programs are associated with a significant increase in structural reforms in all sectors but trade and capital account; estimates statistically significant at the 1% level. However, these raw results are not robust to inclusion of control variables.
- Table 2 (OLS with capacity building and interactions):
  - Individually, IMF programs, training, and technical assistance do not have a significant impact on structural reforms.
  - Interaction Prog × Train is statistically significant at the 5% level for domestic financial sector reforms, capital account, and trade reforms (columns (1), (3), (6) in Table 2).
  - Train variable is binary (above/below median share of officials trained). Quantitative examples:
    - In a country with a 3 year IMF program, receiving above median training (relative to below median) implies an increase in domestic financial sector reforms by 0.099 (that is 9/10 of a standard deviation of domestic financial reform).
    - Example context: median IMF training in a 5 year window corresponds to roughly 200 officials trained (Philippines example).
    - Following an increase in training to above the median:
      - Trade reform would increase by 0.057 (that is 1/2 trade reform standard deviation).
      - Capital account reform would increase by 0.178 (that is 7/10 of capital account reform standard deviation).
  - Interaction Prog × TA and TA individual coefficients are not statistically significant.
  - Interpretation: complementarities between training and IMF programs matter for structural reforms; technical assistance does not show the same complementarities with programs.
- Table 3 (difference-GMM): point estimates similar to Table 2; interaction Prog × Train even more statistically significant when excluding TA (increases sample by about 10 years).
- Table 4 (program failures, OLS): technical assistance is negatively associated with probability of program cancellation; estimates in columns (1) to (6) significant at the 1% level. TA is robustly associated with IMF program implementation.
- Instrumental-variable and robustness checks:
  - Instrument for TA: donor contributions weighted by donors' bilateral aid "country preferences" (donor data from IMF Institute for Capacity Development; bilateral aid data 1981-2010 from OECD, 2012). Predicted TA allocation correlates strongly with actual geographic allocation (Figure 6).
  - Instruments for programs (following Barro and Lee, 2005): (i) United Nations voting patterns - percentage of votes in line with G7/US; (ii) share of IMF Staff from a certain nationality; (iii) a country's IMF quota. The latter two add little explanatory power; instrument strength reduced when including concessional programs.
  - Including a "motivation" control: number of applications a country makes to IMF courses (PATS data). Coefficient on number of applications is positive in all regressions, suggesting it captures authorities' motivation to reform.
  - Table 5 (IV results controlling for applications): coefficient on interaction Prog × Train remains significant — evidence that training in conjunction with programs has a positive effect on reforms.
  - Table 6: technical assistance still does not have a direct impact on structural reform when instrumented.
  - Table 7 (instrumenting TA into program failure regressions): confirm TA helps keep a program on track; effect somewhat smaller than OLS, suggesting both selection (programs likely to succeed get more TA) and causal TA effect on completion.

### Robustness checks and sensitivity
- Results robust to:
  - Using an indicator equal to 1 if country had at least one IMF program over last 5 years (instead of count measure).
  - Splitting training data in three groups rather than two.
  - Splitting world into two regions—advanced Economies and Central and Eastern Europe versus the rest of the world.
  - Results not driven by occurrence of crises.
  - Restricting analysis to concessional programs (results remain robust); concessional programs have stronger reform focus than many non-concessional ones.

### Endogeneity considerations and identification strategies
- Potential biases:
  - Upward bias if an unobserved government willingness to reform drives both capacity building uptake and reforms.
  - Upward bias if active IMF programs lead to more TA and correlate with reforms.
  - Downward bias if "strong" reforming countries do not use IMF capacity building or programs.
- Strategies to address endogeneity:
  - Country fixed effects, time fixed effects, system GMM, instrumenting TA and programs.
  - Instrument for TA based on donor funding weighted by donors' bilateral aid preferences.
  - Instruments for programs: UN voting alignment with G7/US; share of IMF staff from a nationality; IMF quota.
  - Control for number of applications to IMF courses to proxy authorities' motivation.

### Conclusions and policy implications
- Main conclusions:
  - IMF training leads to an increase in structural reforms, but only through IMF programs and only when a significant share of public servants is trained.
  - IMF technical assistance does not significantly lead to more structural reforms, but raises the likelihood of completion of ongoing IMF programs.
  - Evidence of complementarity between IMF programs and IMF capacity building: capacity building increases likelihood that an IMF program provides a window of opportunity for reforms where conditionality and reform ownership reinforce each other.
- Policy recommendations and implications:
  - Consider targeting training more at possible program countries to yield the most impact.
  - Training en masse (reaching a critical mass of officials) seems to yield the most impact, perhaps by fostering a collective culture of reforms.
  - IMF training can help reconcile program conditionality with higher policy ownership by raising awareness and sharing state-of-the-art knowledge.
  - Capacity building may facilitate continued dialogue between authorities and the IMF and influence IMF engagement.

*Source: _wp12232 - Appendix 1 provides a list of existing IMF lending facilities. IMF program data is available (excerpt).*

### REFERENCES

### _wp12232 - REFERENCES

### References
- Alesina, Alberto, Silvia Ardagna & Francesco Trebbi, 2006, “Who Adjusts and When? On the Political Economy of Reforms,” IMF Staff Papers, Mundell-Fleming Lecture, Vol. 53, pp. 1–49.
- Alesina, Alberto & David Dollar, 2000, “Who Gives Foreign Aid to Whom and Why?”, Journal of Economic Growth, Vol. 5, pp. 33–63.
- Arezki, Rabah, Herbert Lui, Marc Quintyn & Frederik Toscani, 2012, “Education Attainment in Public Administrations around the World: Evidence from a New Dataset,” International Monetary Fund, mimeo.
- Ball, Richard & Christopher Johnson, 1996, “Political, Economic and Humanitarian Motivations for PL480 Food Aid: Evidence from Africa,” Economic Development and Cultural Change, Vol. 44, pp. 515–537.
- Barro Robert J. & and Lee Jong-Wha, 2005, “IMF programs: Who is chosen and what are the effects?”, Journal of Monetary Economics, Vol. 52, pp. 1245–1269.
- Barro Robert J. & and Lee Jong-Wha, 2010, "A New Data Set of Educational Attainment in the World, 1950–2010," NBER Working Papers 15902, (Cambridge: National Bureau of Economic Research).
- Blundell, R.W., and S. Bond, 1998, “Initial Conditions and Moment Restrictions in Dynamic Panel Data Models,” Journal of Econometrics, Vol. 87, pp. 115–143.
- Dicks-Mireaux, Louis Mauro and Susan Mecagni & Susan Schadler, 2000, “Evaluating the Effect of IMF Lending to low-income countries”, Journal of Development Economics, Vol. 61, pp. 495–526.
- Drazen, Allen, 2000, Political Economy in Macroeconomics (Princeton: Princeton University Press).
- Drazen, Allen, 2002, Conditionality and Ownership in IMF Lending: A Political Economy Approach, IMF Staff Papers, Vol. 49, pp. 36–67.
- Giuliano, Paola, Prachi Mishra and Antonio Spilimbergo, 2010, “Democracy and Reforms: Evidence from a New Dataset,” IMF Working Paper, WP/10/173 (Washington: International Monetary Fund).
- Ostry, Jonathan D., Alessandro Prati & Antonio Spilimbergo, 2009, “Structural Reforms and Economic Performance in Advanced and Developing Countries,” IMF Occasional Paper no. 268 (Washington: International Monetary Fund).
- Przeworski, Adam & James Raymond Vreeland, 2000, “The Effect of IMF Programs on Economic Growth,” Journal of Development Economics, Vol. 62, pp. 385–421.
- Raghuram G. Rajan & Arvind Subramanian, 2008, "Aid and Growth: What Does the Cross-Country Evidence Really Show?," The Review of Economics and Statistics, November, MIT Press, Vol. 90(4), pp. 643–665.
- Rodrik, Dani, 1996, “Understanding Economic Policy Reform,” Journal of Economic Literature, Vol. 34, pp. 9–41.
- Thacker, Strom, 1999, “The High Politics of IMF Lending,” World Politics, Vol. 52, pp. 38–75.
- Tommasi, Mariano & Andrés Velasco, 1996, “Where Are We in the Political Economy of Reform?,” Journal of Policy Reform, Vol. 1, pp. 187–238.

### Appendix 1. Typology of IMF Programs
- IMF programs can be divided in two main categories depending on whether they are concessional or not. Within those two categories are various sub-categories which are listed below.

- Non-Concessional lending facilities:
  - Stand-by-Arrangements
  - Flexible Credit Line
  - Precautionary and Liquidity Line
  - Extended Fund Facility
  - Rapid Financing Instrument

- Concessional lending facilities:
  - Extended Credit Facility
  - Standby Credit Facility
  - Rapid Credit Facility
  - Poverty Reduction and Growth Facility

- Source: IMF website http://www.imf.org/external/np/exr/facts/eng/list.aspx

*Source: _wp12232 - REFERENCES (PDF).*

### Appendix 2. Table A1. Exploratory Regressions: What country features are associated

### Appendix 2. Table A1. Exploratory Regressions: What country features are associated with Training, Technical Assistance and Programs?

### Determinants of TA, Training per government employee, and Number of program years (Table A1)
- Dependent variables: TA (column 1), Training per gov. empl. (column 2), Programs (column 3). Estimation: ordinary least squares with robust country-clustered standard errors.
- Significant positive associations:
  - Number of program years → TA: 0.476*** (0.155)
  - Crisis Count → TA: 0.202*** (0.0640)
  - Crisis Count → Programs: 0.0885*** (0.0244)
- Significant negative associations:
  - Real GDP pc → TA: -0.000104** (4.71e-05)
  - Real GDP pc → Programs: -6.58e-05*** (1.68e-05)
  - Bureaucratic Quality → Training per gov. empl.: -0.0332* (0.0194)
  - Bureaucratic Quality → Programs: -0.346** (0.163)
  - Inflation → Programs: -0.000716** (0.000325)
- Other notable coefficients (not significant at conventional levels unless marked):
  - Foreign Debt/GDP → TA: 0.005190 (0.00534)
  - Foreign Debt/GDP → Training per gov. empl.: 0.000392 (0.000274)
  - Foreign Debt/GDP → Programs: 0.00381* (0.00199)
  - Polity 2 Score → Programs: 0.0671** (0.0311)
  - Corruption → TA: -0.274 (0.367)
  - Average Years of Schooling → TA: 0.264 (0.207)
- Sample and controls:
  - Time Fixed Effects: Yes; Constant: Yes.
  - Observations: TA: 273; Training per gov. empl.: 222; Programs: 273.
  - Number of ifs: 927 / 592 / 273 (aligned with columns 1/2/3).

### Program association with structural reforms (Table 1)
- Dependent variables: indices of reform in six sectors (Domestic Finance, Current Account, Trade Reforms, Agriculture, Networks, Capital Account).
- Number of years with an active program is positively associated with reforms in:
  - Domestic Finance: 0.0215*** (0.00387)
  - Current Account: 0.0127*** (0.00344)
  - Agriculture: 0.0245*** (0.00330)
  - Networks: 0.00770*** (0.00225)
- Estimation: FE; Observations vary by sector (477 / 921 / 765 / 1,049 / 1,157 / 477).

### Training and TA interacted with programs — OLS results (Table 2)
- Key coefficients:
  - Interaction lagged Training and contemp. Program:
    - Domestic Finance: 0.0156** (0.00757)
    - Trade Reforms: 0.0191** (0.00774)
    - Capital Reforms: 0.0595** (0.0254)
  - TA Person-Years:
    - Domestic Finance: -0.00782* (0.00429)
    - Trade Reforms: 0.00717* (0.00410)
  - Number of years with an active program:
    - Domestic Finance: -0.0174*** (0.00601)
    - Networks: -0.0270* (0.0152)
- Controls: Polity 2 Score, Average years of schooling, Bureaucratic Quality. Country and Time Fixed Effects included.
- Observations by column: 205 / 241 / 245 / 215 / 232 / 205. R-squareds reported per column.

### Training and TA interacted with programs — difference-GMM (Table 3)
- Key coefficients:
  - Training per gov. empl. above the median lagged:
    - Current Account: 0.0887** (0.0410)
    - Agriculture: 0.118* (0.0697)
    - Networks: 0.174* (0.0954)
  - Interaction lagged Training and contemp. Program:
    - Domestic Finance: 0.0136* (0.00705)
    - Trade Reforms: 0.00829** (0.00419)
    - Capital Reforms: 0.0643* (0.0365)
  - TA-Person Years:
    - Networks: -0.0207*** (0.00748)
- Estimation: Two-Step difference-GMM; Robust S.E. Observations per column: 121 / 148 / 138 / 131 / 151 / 121.

### Capacity building and program failure (Table 4)
- Dependent variable: ratio of cancelled programs over total programs (OLS with country and time fixed effects).
- Total Technical Assistance in the last 5 years is associated with fewer cancelled programs:
  - Column examples: -0.0142*** (0.00338); -0.0208*** (0.00566); -0.0233*** (0.00627)
- Total Training per gov. empl. in the last 5 years:
  - Example: -0.0360* (0.0218) in column (1) but mixed significance across specifications.
- Lagged Government Debt/GDP:
  - Positive association with cancellation ratio in some specs: 0.00330** (0.00147); 0.00422*** (0.00151)
- Estimation: OLS, country and time fixed effects. Observations vary widely across specs (629 / 604 / 629 / 163 / 476 / 123).

### Impact of training on structural reforms controlling for country motivation — system-GMM (Table 5)
- Key coefficients:
  - Interaction lagged Training and contemp. Program:
    - Domestic Finance: 0.0207* (0.0108)
    - Networks: 0.0590** (0.0230)
    - Capital: 0.0401* (0.0239)
  - Number of applications lagged:
    - Trade Reforms: 0.00377** (0.00192)
  - Polity 2 Score and Average years of schooling show several positive and significant associations across sectors.
- Estimation: System and Difference GMM. Observations per column: 182 / 218 / 222 / 192 / 209 / 182. Hansen Test p-values reported per column.

### Impact of technical assistance on structural reforms using instrument based on donor funding — system-GMM (Table 6)
- Key coefficients:
  - TA Person-Years:
    - Domestic Finance: -0.0109** (0.00517)
  - Interaction TA and Program:
    - Domestic Finance: 0.00237** (0.00119)
    - Networks: 0.00199** (0.00161)
  - Number of applications lagged:
    - Current Account: 0.0243** (0.0116)
- Estimation: System and Difference GMM with donor-funding-based instrument. Observations per column: 182 / 218 / 222 / 192 / 209 / 182. Hansen Test p-values reported.

### Capacity building and program success using instrument based on donor financing — IV (Table 7)
- Dependent variable: cancelled programs / total programs. Method: IV (TATATATA instrumented; Donor Contr. as excluded instrument). Country and time fixed effects included.
- Total Technical Assistance in the last 5 years:
  - Negative association with cancelled program ratio in several IV specifications:
    - Column examples: -0.0374* (0.0205); -0.0390* (0.0230); -0.0382** (0.0191)
  - Some columns show unstable precision (e.g., col (6) coefficient 0.0691 with s.e. 0.329).
- Total Training per gov. empl. in the last 5 years: mixed and generally insignificant in IV specs (examples: -0.0198 (0.0292); 0.0748* (0.0442) in column (4)).
- Other controls: Lagged Government Debt/GDP: 0.00468** (0.00194) in some specs.
- Observations by column: 558 / 533 / 558 / 137 / 420 / 106.

*Source: _wp12232 - Appendix 2. Table A1. Exploratory Regressions: What country features are associated with Training, Technical Assistance and Programs?*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12232.pdf_