## _wp05126

## Source details

**Canonical URL:** [_wp05126](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05126.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05126.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05126.pdf.json)

---

### Research question and motivation
- Central question: Why does foreign aid often fail to produce robust long-run growth effects despite immediate resource needs (e.g., textbooks, medicines, roads)?
- Context from recent literature:
  - Easterly (2003); Easterly, Levine, and Roodman (2004); Hansen and Tarp (2001); Roodman (2004); Rajan and Subramanian (2005) — studies suggesting limited aid-growth association even with good policies.
  - Clemens, Radelet, and Bhavnani (2004) — short-term impacts for certain aid types but not discernible longer-term effects.
  - Roodman (2004) quoted conclusion: “if there is one strong conclusion from this literature, it is that on average aid works well outside the tropics but not in them.”

### Hypothesized adverse channels of aid
- Institutional weakening:
  - Aid may reduce incentives for domestic revenue mobilization (tax effort), creating dependency (Azam, Devarajan, and O’Connell, 1997; Adam and O’Connell, 1999).
  - Expanded government resources may relax accountability, fostering corruption and weakening institutions over the long run.
- Competitiveness effects (macroeconomic “Dutch Disease”):
  - Resource-price channel: aid spending bids up prices of critical domestic resources (e.g., fees to contractors; salaries to engineers, doctors, teachers, civil servants, aid administrators), allowing non-traded industries to raise prices while tradables cannot.
  - Exchange-rate channel:
    - Flexible exchange regime: aid inflows can push up the nominal exchange rate when the central bank sells aid inflows in the domestic foreign exchange market, reducing traded-sector competitiveness if wages do not fall.
    - Fixed exchange regime: aid-driven domestic demand increases prices of scarce domestic resources, reducing competitiveness of internationally exposed industries.
  - Common outcome: real exchange rate appreciation and a shrinking traded sector.

### Methodology and empirical strategy
- Cross-country cross-industry framework following Rajan and Zingales (1998):
  - Test whether industries most exposed to the wage/price/exchange-rate channels grow relatively slower in countries with higher aid inflows.
  - Exploits within-country differential effects and country-level treatment (aid inflows).
  - Controls: country and industry fixed effects.
- Labor-intensity measure: average across sample countries of the share of labor compensation to value-added in each industry.
- Core estimating equation (paraphrased):
  - Growth_ij = Constant + country fixed effects + industry fixed effects + (initial industry share) + α*(Aid/GDP_j * Labor Share_i) + ε_ij
  - Expectation: α negative.
- Instrumentation: predicted aid constructed from donor-recipient strategic, historic, and cultural links to address endogeneity.

### Data and sample
- Primary sources: UNIDO Industrial Statistics Database (2003); OECD DAC; WITS; World Development Indicators; IMF IFS.
- Coverage and sample construction:
  - UNIDO 3-digit ISIC Revision 2 data for 1980s and 1990s.
  - Include sectoral observations with average growth computed over at least a seven-year period in the decade.
  - Final sample: 33 countries for the 1980s and 15 countries for the 1990s; 28 industries present in UNIDO for these countries.
- Key summary statistics (Table 1):
  - Median growth rate of value added for industries: 1.5 percent for 1980-90 and 3.7 percent for 1990-2000.
  - Average aid inflow: 5.8 percent of GDP for the 33 countries in 1980-1990 sample; 5.0 percent of GDP for the 15 countries in 1990-2000 sample.
  - Average labor intensity of industries: about 40 percent in the 1980s and 36 percent in the 1990s.
  - Sample sizes referenced in data construction: 954 observations for forty countries (1980s) and 642 observations for twenty eight countries (1990s) in intermediate steps; final regression observations reported elsewhere.

### Main empirical findings
- Aid undermines competitiveness of labor-intensive and exportable industries:
  - Interaction Aid/GDP(j) * Labor share(i) is consistently negative and statistically significant at the 5 percent level in core specifications for both decades.
  - In higher-aid countries, labor-intensive sectors grow slower relative to capital-intensive sectors; exportable sectors grow slower relative to non-exportable sectors.
  - Consequences: slower employment growth in affected sectors and lower sectoral shares.
- IV versus OLS:
  - IV coefficient larger (in absolute value) than OLS, suggesting measurement error attenuation and/or endogeneity; core result robust to instrumentation.
- Magnitude example (core IV specification, column 3 Table 2A):
  - A one standard deviation increase in aid in country j (about 4.2 percentage points in aid/GDP) combined with a one standard deviation increase in labor intensity (about 8.4 percentage points in labor share) yields large relative effects: e.g., apparel sector should grow 3.5 percent per year slower relative to glass and products in Zambia than in Honduras.
  - Average growth rate in the sample is 1.8 percent.
- First-stage instrument strength:
  - F-statistic = 23.5 for the 1980s; F-statistic = 27.5 for the 1990s.

### Non-parametric and robustness evidence
- Non-parametric depiction (Chart 2):
  - Industries split into above- and below-median labor share groups; country-level difference in growth plotted against Aid/GDP shows a negative relationship; no single country drives the relationship.
  - Chart 1A coefficient (1980s): coef = -.03191105, (robust) se = .01141765, t = -2.79.
  - Chart 1B coefficient (1990s): coef = -.01950493, (robust) se = .00873422, t = -2.23.
- Robustness checks (samples, outliers, winsorization, country deletions):
  - 1980s IV examples:
    - Outliers excluded: -0.032** [0.015], observations 683.
    - Winsorize at 1%/99%: -0.031** [0.013], observations 712.
    - Winsorize at 5%/95%: -0.023** [0.010], observations 712.
    - Exclude Israel, Poland, Thailand: -0.028* [0.014], observations 658.
  - 1990s IV examples:
    - Outliers excluded: -0.020** [0.010], observations 338.
    - Winsorize at 1%/99%: -0.022** [0.009], observations 357.
    - Winsorize at 5%/95%: -0.015* [0.008], observations 357.
  - Deletion tests: 60 sub-samples (1980s) and 43 sub-samples (1990s); coefficients statistically significant in all cases.
  - Alternative instruments (e.g., donor budgetary position) produce qualitatively similar results.
  - Clustering standard errors by industry and by country leaves coefficients significant.

### Different aid measures and channels
- Types of aid (Table 5A/5B):
  - Total aid excluding technical assistance: coefficient larger in magnitude: -4.1 percent (compared to -3.5 percent core).
  - Technical assistance alone: coefficient = -2.6 percent (negative and significant).
  - Short-term and long-term impact categorizations (Clemens et al. database): negative and significant interactions in both short-term and long-term columns.
- Labor versus capital channel (Table 6):
  - Aid/GDP * Financial dependence(i) coefficients not statistically significant.
  - Aid/GDP * Labor share(i) coefficients statistically significant and similar in magnitude to core results.
  - Interpretation: evidence favors an input-price / wage-competitiveness channel rather than a capital-relief channel.

### Exportability and alternative competitiveness indicators
- Exportability measures (Table 7):
  - Exportability1 (dummy above-median exports/value added) and Exportability2 (average exports/value added) both show negative Aid * Exportability interactions.
  - 1980s: Aid/GDP*Exportability1 = -0.052** [0.023]; Aid/GDP*Exportability2 = -0.077 [0.051].
  - 1990s: Aid/GDP*Exportability1 = -0.047** [0.020]; Aid/GDP*Exportability2 = -0.038* [0.021].
  - Correlation between labor-intensity and exportability = 0.34.
  - Implication: aid negatively affects export sectors, confirming core labor-intensity result.

### Values vs quantities; nominal vs real measures; employment
- Real value-added proxy (deflation by per person nominal wage growth) and employment growth outcomes (Tables 8A/8B):
  - Interaction coefficients maintain predicted negative sign and statistical significance; magnitudes close to core specification.
  - Employment growth interaction negative and significant for the 1980s; 1990s results marginally weaker but mostly negative and significant.
  - Examples (Table 8A, 1980s):
    - Aid/GDP*Labor share on deflated value added: -0.029** [0.012] (col (1)); -0.021*** [0.008] (col (2)).
    - Initial employment(ij): -0.007** [0.003] (col (4)).
  - Examples (Table 8B, 1990s):
    - Aid/GDP*Labor share: -0.016* [0.010] (col (1)); -0.012* [0.007] (col (2)).
    - Aid/GDP*Exportability1 on employment: -0.053** [0.022] (col (3)).

### Aid, overvaluation, and causality
- Overvaluation measure: residual from regression of real exchange rate on PPP per capita income (Balassa-Samuelson relationship) averaged over end decades.
- Correlation and IV evidence:
  - Aid inflows and real exchange rate overvaluation are positively correlated; correlation strengthens over time (Charts 3A–3C).
  - First-stage regressions of overvaluation on instruments: coefficient on instruments = .283 (1980s) with t = 4.31; .369 (1990s) with t = 6.69.
  - IV second-stage using predicted overvaluation interacted with labor intensity: the exogenous (aid-related) component of overvaluation hurts competitiveness in the same pattern as exogenous aid.
  - Comparison: a 1 percentage point increase in aid/GDP has the same impact on labor-intensive sectors as a 3.8 percent overvaluation (as noted in text).
- Table 9 examples:
  - OLS (1980s): Overvaluation(j)*Labor Share(i) = -0.017*** [0.004], Observations 702, R-squared 0.43.
  - IV (1980s): Overvaluation*Labor Share = -0.077** [0.035], Observations 702, R-squared 0.20.
  - OLS (1990s): Overvaluation*Labor Share = -0.017*** [0.006], Observations 357, R-squared 0.33.
  - IV (1990s): Overvaluation*Labor Share = -0.034** [0.017], Observations 357, R-squared 0.31.

### Channels of influence: skilled and unskilled wages
- Wage channel evidence (1980s):
  - Using U.S. skill indices (Rajan and Wulf, 2004) as proxies for industry skill requirements, regression of annual average growth of dollar wages per employee on skill-index * aid interaction yields:
    - Interaction coefficient = 0.10, standard error = 0.004, significant at the 1 percent level.
  - Interpretation: wages in sectors using relatively scarce skilled labor grew faster in countries receiving more aid, consistent with the wage-pressure channel.

### Time pattern: 1980s versus 1990s
- Adverse competitiveness effects stronger in the 1980s:
  - Core specification: 1990s coefficient about two-thirds that of 1980s, while average value added growth almost twice as high in the 1990s (average growth rates: 0.018 in 1980s, 0.030 in 1990s in Table 2A note).
- Possible mitigating developments in the 1990s:
  - Greater trade openness and exchange-rate flexibility.
  - Improved macroeconomic management of aid inflows or better utilization of aid.
- Nonetheless, negative effects persist in the 1990s but at smaller magnitude.

### Remittances: contrasting effects with aid
- Remittances (1990s analysis only due to data reliability):
  - None of the remittance interaction coefficients are statistically significant (Table 10).
  - Examples (Table 10): Remittances(j)*Labor share(i) = 0.004 [0.006] (col (4)); Remittances(j)*Exportability index(i) = 0.006 [0.013] (cols (4)-(5)); Observations 343.
- Patterns and explanations:
  - Remittances do not lead to higher wage growth in industries more skilled-labor intensive (estimates not reported).
  - Charts 4A/4B: countries with overvalued exchange rates in the 1990s received significantly lower remittances; remittances for 1990-2000 plotted against 1980-90 overvaluation show negative association.
  - Endogeneity interpretation: emigrants may reduce remittances when exchange rates are overvalued or when exchange controls exist; remittances tend to dry up when overvaluation occurs, which helps explain why remittances do not generate the same adverse competitiveness effects as aid.

### Broader implications, externalities, and distributional concerns
- Documented adverse side effects:
  - Systematic adverse effects on growth, wages, and employment in labor-intensive and export sectors.
  - Labor-intensive industries are important for employment generation and absorbing surplus agricultural labor; their shrinkage can affect inequality and poverty.
  - Aid-induced shifts away from export-oriented labor-intensive manufacturing may forgo productivity spillovers, learning, and constituencies that support good governance and macroeconomic management.
- Illustrative cross-country pattern (Chart 5):
  - Concentration measure: ratio of value added share of above-median labor intensity industries to below-median industries.
  - Plot of per capita growth 1980–2000 against 1990 concentration shows strong positive association: emerging Asia and South Asia concentrations about 0.6 and associated with higher growth; aid-dependent Africa concentration about 0.47 and associated with lower growth.

### Policy-relevant conclusions and recommendations
- Do not assume increased aid is harmless; advocates must demonstrate net long-run benefits, not just absence of harm.
- Key mitigation priorities:
  - Careful macroeconomic management to prevent real exchange rate overvaluation.
  - Consider absorptive capacity limits when scaling up aid.
  - Design aid delivery and spending modalities to minimize demand pressure on scarce domestic resources (e.g., skilled labor, tradable-sector inputs).
  - Build supply-side capacity (e.g., expand skilled labor supply) before large-scale aid absorption; phase-in aid increases.
  - Consider sterilization and reserve accumulation options to mitigate real appreciation while weighing costs (e.g., higher domestic interest rates, crowding out).
  - Encourage donors to reduce impediments to exports and recipients to reduce trade barriers so trade can help absorb aid.
  - Adopt experimental approaches to aid delivery with attention to delivery mechanisms, incentives, and spillovers.
- Trade-offs noted:
  - Sterilization/reserve accumulation mitigate real appreciation but may reduce immediate development spending impact.
  - Massive short-term expansions in education/training may create short-run adverse effects; recommend paced approaches.
- Research implication:
  - Demonstrating adverse channels raises the bar for claims that increased aid will produce sustained growth; program design must address these channels explicitly.

### Summary conclusions
- Evidence supports a credible causal channel: aid inflows cause real exchange rate appreciation and harm labor-intensive and exportable industries via wage and price channels.
- Methodology measures relative sectoral growth effects; the net effect of aid on aggregate growth remains an open question and is sensitive to specification and horizon.
- Findings emphasize the need for careful macroeconomic and policy design to avoid adverse competitiveness side effects and to ensure aid contributes to sustained growth.

*Source: _wp05126 - IMF working paper excerpt.*

### 1. Summary Statistics...................................................................................................

### _wp05126 - 1. Summary Statistics...................................................................................................

### Research question and motivation
- Central question: Why does foreign aid often fail to produce robust long-run growth effects despite immediate resource needs (e.g., textbooks, medicines, roads)?
- Context from recent literature:
  - Studies suggesting limited aid-growth association even with good policies: Easterly (2003), Easterly, Levine, and Roodman (2004), Hansen and Tarp (2001), Roodman (2004), Rajan and Subramanian (2005).
  - Clemens, Radelet, and Bhavnani (2004) find short-term impacts for certain aid types but not discernible longer-term effects.
  - Roodman (2004) conclusion quoted: “if there is one strong conclusion from this literature, it is that on average aid works well outside the tropics but not in them.”

### Hypothesized adverse channels of aid
- Two broad adverse mechanisms are emphasized:
  - Institutional weakening:
    - Aid may reduce incentives for domestic revenue mobilization (tax effort), creating dependency (Azam, Devarajan, and O’Connell, 1997; Adam and O’Connell, 1999).
    - Expanded government resources may relax accountability, potentially fostering corruption and weakening institutions over the long run.
  - Competitiveness effects (macroeconomic “Dutch Disease”):
    - Flexible exchange rate channel: Aid inflows can push up the nominal exchange rate, reducing competitiveness of traded goods if wages in traded sector do not fall.
    - Fixed exchange rate channel: Aid-driven domestic demand increases prices of scarce domestic resources (e.g., skilled workers, coastal land), reducing competitiveness of internationally exposed industries.
    - Common outcome: real exchange rate overvaluation causing adverse consequences for growth of the traded goods sector.

### Methodology and empirical strategy
- Cross-country cross-industry framework following Rajan and Zingales (1998) logic:
  - Test whether industries most exposed to the hypothesized channel (e.g., labor-intensive industries facing higher wages) grow relatively slower in countries with higher aid inflows.
  - Exploits within-country differential effects (labor-intensive vs. non-labor-intensive growth) and country-level treatment (aid inflows).
  - Controls for country and industry fixed effects to reduce sensitivity to aggregate cross-country regression specifications.

### Main empirical findings
- Evidence consistent with aid undermining competitiveness:
  - In higher-aid countries, labor-intensive sectors grow slower relative to capital-intensive sectors.
  - Exportable sectors grow slower relative to non-exportable sectors in higher-aid countries.
  - Reduced competitiveness is associated with slower employment growth in affected sectors and a lower share of these sectors in the economy.
- Aid, overvaluation, and causality:
  - Aid and real exchange rate overvaluation are positively correlated across countries, with the relationship strengthening over time.
  - Overvaluation correlates with exogenous determinants of aid, suggesting aid contributes to overvaluation.
  - The exogenous (aid-related) component of overvaluation induces the same relative growth patterns across sectors as the exogenous component of aid.
  - Additional finding: aid causes some adverse effects through increases in wages of skilled workers (one source of overvaluation).
- Remittances:
  - Remittances and aid share the characteristic of being “unrequited” transfers, but remittances do not appear to have the same adverse effects on competitiveness.
  - Possible reasons:
    - Different spending patterns (e.g., more on hiring unskilled labor or housing extensions) that do not increase demand for scarce skilled resources.
    - Endogenous response: remittances are far smaller to countries with overvalued exchange rates; senders may avoid remitting to overvalued/exchange-restricted economies or prefer sending goods when real value is diminished.

### Broader implications and interpretation
- Traded goods sector importance:
  - The traded goods sector is typically the source of productivity improvements, learning-by-doing spillovers, and scarce foreign exchange earnings.
  - Adverse impacts of aid on this sector could retard overall economic growth and reduce constituencies that support good governance and macroeconomic management.
  - Empirical note: developing countries with a greater share of manufacturing in the most labor-intensive sectors seem to have had better growth performance.
- Time dynamics:
  - Negative competitiveness effects operate over time: squeezing profitability, reducing new entry, and eventually causing firm closures and sectoral shrinkage.
  - Reduced competitiveness can increase dependence on future aid, reinforcing a dependency cycle.

### Policy-relevant conclusions and recommendations
- Do not assume that increased aid is harmless; advocates for more aid should be convinced it does net good, not merely that it does no harm.
- Key policy priorities to mitigate adverse effects:
  - Pay careful attention to macroeconomic management to prevent real exchange rate overvaluation.
  - Consider absorptive capacity limits when scaling up aid.
  - Design aid delivery and spending modalities to minimize demand pressure on scarce domestic resources (e.g., skilled labor, tradable-sector inputs).
  - Recognize that humanitarian aid remains vital for crisis relief and immediate welfare objectives; its purpose differs from long-run growth objectives.
- Research implication:
  - Establishing existence of adverse channels raises the bar for claims that increased aid will produce sustained growth; program design must address these channels explicitly.

*Source: _wp05126 - 1. Summary Statistics (IMF working paper excerpt).*

### section IV, we offer some concluding remarks.

### _wp05126 - section IV, we offer some concluding remarks.

### A. Theory
- Focus: aid could hurt growth by reducing competitiveness of the traded goods sector and reallocating resources toward the non-traded sector.
- Two channels described:
  - Resource-price channel: aid spending bids up prices of critical resources common to traded and non-traded industries (fees to contractors; salaries to engineers, doctors, teachers, civil servants, aid administrators). Non-traded industries can raise output prices; tradable sector cannot and loses competitiveness and profitability.
  - Exchange-rate channel: in a flexible exchange regime, aid inflows may push up the nominal exchange rate (e.g., when the central bank sells aid inflows in the domestic foreign exchange market), rendering the traded goods sector uncompetitive if wages do not adjust downward.
- Both channels lead to a real exchange rate appreciation and a shrinking traded sector; the inflationary environment for wages and prices depends on the exchange rate regime.
- Mitigating factors: aid spent on traded goods (imported capital goods, foreign consultants), aid spent on goods not in limited supply (unskilled labor), and aid accompanied by domestic fiscal adjustment reduce the likelihood of excessive wage pressure and real exchange rate appreciation.

### B. Empirical strategy
- Methodology: follows Rajan and Zingales (1998) to test whether aid reduces competitiveness of tradable sectors by exploiting cross-industry differences within countries.
- Core idea: industries most affected by the wage-pressure channel should grow differentially in countries where the channel is operative (countries receiving more aid).
- Due to data limits (sectoral data only for manufacturing, largely tradable), the authors exploit variation in industry labor costs rather than tradability across sectors.
- Labor-intensity measure: average across sample countries of the share of labor compensation to value-added in each industry.
- Robustness check: construct an “exportability” proxy for a sector and test whether more exportable sectors grow more slowly in higher-aid countries; labor-intensity preferred as it is less prone to endogeneity bias.
- Outcome measure: annual average rate of growth of value added of industry i in country j over a ten-year period, normalizing nominal value added growth by the GDP deflator; country fixed effects included.
- Estimation equation (paraphrased):
  - Growth_ij = Constant + country fixed effects + industry fixed effects + (initial industry share) + α*(Aid to country j * Labor Intensity of industry i) + ε_ij
  - Expectation: α negative (countries receiving more aid see a more negative impact in labor-intensive industries).
- Advantages: controlling for country and industry fixed effects reduces omitted variable bias; within-country industry differentials less sensitive to endogenous targeting of aid.
- Instrumentation: instrument aid using strategic, historic, and cultural links between donor and recipient to disentangle aid effects from policy-driven exchange rate and trade effects.

### C. Data and their sources
- Industry value added growth: Industrial Statistics Database (2003) of UNIDO; data at 3-digit ISIC Revision 2; available for the 1980s and 1990s.
- Sample construction:
  - Include sectoral observations with average growth computed over at least a seven-year period in the decade.
  - UNIDO database coverage: 47 developing countries for the 1980s and 31 countries for the 1990s (before sample restrictions).
  - Exclude richer emerging market countries (objective criterion) and focus on poorer ones; final sample comprises countries that receive aid greater than 1 percent of GDP or are low-income countries according to World Bank definitions in the initial year.
  - Final sample: 33 countries for the 1980s and 15 countries for the 1990s.
  - UNIDO database contains data on 28 industries in these countries.
- Summary statistics (Table 1 reported in source):
  - Median growth rate of value added for industries: 1.5 percent for 1980-90 and 3.7 percent for 1990-2000.
  - Average aid inflow: 5.8 percent of GDP for the 33 countries in 1980-1990 sample; 5.0 percent of GDP for the 15 countries in 1990-2000 sample.
  - Average labor intensity of industries: about 40 percent in the 1980s and 36 percent in the 1990s.

### D. Results for Aid and Labor Intensity
- Core specifications reported (Table 2A):
  - Columns 1-3 report estimates for the 1980s; columns 4-6 for the 1990s.
  - Column 1: OLS estimate; Column 2: “reduced form” OLS with the instrument entered directly; Column 3: “pure” IV specification.
- Main finding:
  - In all specifications, the interaction term (Aid * Labor Intensity) is negative and statistically significant at the 5 percent confidence level.
  - Interpretation: value added by labor-intensive industries grows relatively more slowly than for other industries in countries that receive more aid.
- IV vs OLS:
  - The IV coefficient is larger (in absolute value) than the OLS coefficient, suggesting measurement error (attenuation bias) may be important in OLS as well as endogeneity.
  - The core result does not depend on instrumentation, although IV increases the measured magnitude of the competitiveness effect.
- Results for the 1990s are broadly similar to those for the 1980s.

*Source: _wp05126 - section IV, we offer some concluding remarks.*

### Appendix 1 lists all countries for which data were available as well as those countries that

### _wp05126 - Appendix 1 lists all countries for which data were available as well as those countries that

### Core result and magnitudes
- The core IV specification (column 3, Table 2A) implies that a one standard deviation increase in aid in country j (about 4.2 percentage points in aid/GDP) combined with a one standard deviation increase in the labor intensity of an industry (about 8.4 percentage points in labor share) yields large relative effects:
  - Example: apparel sector should grow 3.5 percent per year slower relative to glass and products in Zambia than in Honduras.
  - Average growth rate in the sample is 1.8 percent.
- All regressors are normalized so coefficients measure the impact of a one standard deviation increase in the j variable times a one standard deviation increase in the i variable.
- First-stage instruments are precisely estimated and strong:
  - F-statistic of 23.5 for the 1980s.
  - F-statistic of 27.5 for the 1990s.
- Table 2A estimates (selected):
  - Initial industry share(ij): -0.026*** (1980s OLS) with [0.005] standard error.
  - Aid/GDP(j)*Labor share(i): -0.035** (1980s IV, column 3) with [0.014] standard error.
  - Observations: 712 (1980s), 357 (1990s).
  - R-squared: 0.41–0.38 (1980s), 0.33–0.32 (1990s).
  - Average growth rates reported: 0.018 (1980s) and 0.030 (1990s) in Table 2A note.

### Non-parametric visualization and difference-in-difference
- Alternative non-parametric approach (Chart 2):
  - Industries split into above- and below-median labor share groups.
  - For each country, difference in average growth between above- and below-median industries is plotted against aid/GDP.
  - Result: difference in growth is negatively related to aid received; no single country drives the relationship.

### Robustness checks: samples, outliers, instrumentation, clustering
- Robustness to sample manipulations (Tables 3A and 3B):
  - 1980s IV estimates (examples):
    - Outliers excluded: Aid/GDP*Labor share = -0.032** [0.015], observations 683.
    - Winsorize at 1%/99%: -0.031** [0.013], observations 712.
    - Winsorize at 5%/95%: -0.023** [0.010], observations 712.
    - Exclude Israel, Poland, Thailand: -0.028* [0.014], observations 658.
  - 1990s IV estimates (examples):
    - Outliers excluded: -0.020** [0.010], observations 338.
    - Winsorize at 1%/99%: -0.022** [0.009], observations 357.
    - Winsorize at 5%/95%: -0.015* [0.008], observations 357.
- Deletion tests:
  - For the 1980s: checked 60 different sub-samples (delete one country/industry); coefficients statistically significant in all 60.
  - For the 1990s: checked 43 sub-samples; coefficients statistically significant in all 43.
  - Total 103 cases with coefficients statistically significant and broadly unchanged.
- Instrumentation robustness:
  - Rajan and Subramanian (2005) second set of instruments (donor budgetary position) yields qualitatively similar results.
- Clustering:
  - Columns 4 and 5 present standard errors clustered by industry and by country; coefficient estimates remain significant in both decades.

### Different types of aid (Table 5A and 5B)
- Comparing total aid vs technical assistance:
  - Total aid without technical assistance: coefficient in core specification is larger in magnitude: -4.1 percent (compared to -3.5 percent core).
  - Technical assistance alone: coefficient is -2.6 percent (still negative and significant).
- Short-term vs long-term aid (Clemens et al. database):
  - Aid likely to have short-term impact (column 3) and long-term impact (column 4): interaction estimates are negative and statistically significant in both cases.
- Interpretation:
  - All forms of aid show the robust adverse competitiveness effect, though magnitudes vary; fungibility and proxies may explain persistence of negative coefficients for technical assistance.

### Labor versus capital channel (Table 6)
- Tested whether aid favors capital-dependent industries using Rajan and Zingales (1998) financial dependence variable:
  - If aid eased financing constraints, the Aid * Financial Dependence interaction should be positive.
  - Results: Only the labor intensity interaction is statistically significant; financial dependence interaction is not.
  - Magnitudes of labor intensity-aid interaction similar to those in columns 3 and 6 of Table 2A.
- Conclusion: evidence points away from capital-relief channel and towards input-price / wage-competitiveness channel.

### Alternative competitiveness indicator: exportability (Table 7)
- Two exportability proxies:
  - exportability1: indicator = 1 if exports to value added (averaged across developing countries) > median across industries.
  - exportability2: ratio of exports to value added averaged across developing countries.
  - Correlation between labor-intensity and exportability measure is 0.34.
- Results:
  - Exportable industries grew slower in countries receiving more aid in both 1980s and 1990s (Table 7).
  - Three out of four coefficients statistically significant at standard levels; the fourth close to significant.
- Implication: aid negatively affects export sectors, confirming core labor-intensity result.

### Values vs quantities; nominal vs real measures (Table 8A and 8B)
- Real value-added proxy using industry-specific price deflator approximated by per person nominal wage growth:
  - Dependent variable: nominal value added growth deflated by per person nominal wage growth in the industry.
  - Interaction coefficients predicted sign and significant; magnitudes close to core specification.
- Employment growth (columns 4–6, Table 8A/8B):
  - Dependent variable: average employment growth in the decade.
  - Aid interaction term has negative sign and is significant for the 1980s.
  - 1990s results are marginally weaker; most interaction coefficients negative and significant, except interaction with exportability2 in 1990s.
- Conclusion: both nominal and adjusted real measures and employment growth point to lower growth in labor-intensive and exportable sectors when aid is higher.

### Aid, overvaluation, and causality (Tables 9 and first-stage statistics)
- Overvaluation measure constructed via Frankel (2004) method: residual of regression of real exchange rate on PPP per capita income (Balassa-Samuelson relationship).
  - Average overvaluation computed as average of values for end decades (e.g., average of 1980 and 1990 for 1980-90).
- Plots (Figures 3A–3C): aid inflows and overvaluation are positively correlated; correlation strengthens over time.
- IV evidence linking aid to overvaluation:
  - First-stage regressions of overvaluation on instruments: coefficient on instruments .283 (1980s) and .369 (1990s) with t-values 4.31 and 6.69.
  - IV second-stage (using predicted overvaluation interacted with labor intensity) suggests the component of overvaluation caused by exogenous aid inflows hurts competitiveness.
  - Comparison: a 1 percentage point increase in aid/GDP has the same impact on labor-intensive sectors as a 3.8 percent overvaluation (as noted in text).

### Channels of influence: skilled and unskilled wages (1980s evidence)
- Wage channel hypothesis:
  - Aid-driven spending raises prices in non-traded sector; wages for scarce worker-types (e.g., skilled labor) rise, squeezing traded-sector profitability.
  - Many exchange rates were largely fixed in the 1980s, so nominal appreciation channel limited; wage channel should be observable.
- Empirical check using U.S. skill indices (Rajan and Wulf, 2004) as proxy for industry skill requirements:
  - Regression of annual average growth of dollar wages per employee in sector-country over the 1980s on interaction between skill index and aid.
  - Interaction coefficient = 0.10 (standard error 0.004) and significant at the 1 percent level.
  - Interpretation: wages in sectors using relatively scarce skilled labor grew faster in countries receiving more aid, consistent with the wage channel for overvaluation.

### Time pattern: 1980s versus 1990s
- Adverse competitiveness effects stronger in magnitude and statistical significance in the 1980s:
  - Core specification: 1990s coefficient about two-thirds that of 1980s, while average value added growth almost twice as high in the 1990s.
- Possible explanations:
  - Greater trade openness and exchange-rate flexibility in the 1990s may have attenuated measured effects.
  - Improved macroeconomic management of aid inflows or better utilization of aid in the 1990s.
- Nonetheless, adverse effects remain present in the 1990s, albeit smaller.

### Remittances: contrasting effects with aid (Table 10 and Charts 4A–4B)
- Remittances analysis restricted to the 1990s (reliable data only for 1990s).
- Table 10 (columns 1–5): none of the remittance interaction coefficients are statistically significant.
- Patterns and explanations:
  - Micro-evidence suggests remittances can positively affect entrepreneurship, labor supply, and investment.
  - Remittances do not lead to higher wage growth in industries that are more skilled-labor intensive (estimates not reported).
  - Chart 4A and 4B: countries with overvalued exchange rates in the 1990s received significantly lower remittances; similar pattern plotting remittances in the 1990s against 1980s overvaluation.
  - Endogeneity interpretation: emigrants may reduce remittances when exchange rates are overvalued or when exchange controls/dual rates exist; thus remittances tend to dry up when overvaluation occurs, implying remittances do not generate the same adverse competitiveness effects as aid.

### Broader implications, externalities, and distributional concerns
- Adverse side effects of aid inflows documented:
  - Systematic adverse effects on growth, wages, and employment in labor-intensive and export sectors.
  - Labor-intensive industries are important for employment generation and absorbing surplus agricultural labor; their shrinkage can affect inequality and poverty.
  - Aid-induced shifts away from export-oriented labor-intensive manufacturing may forgo productivity spillovers, learning, and political-economic support for sensible policy.
- Illustrative cross-country pattern (Chart 5):
  - Ratio of share of value added of above-median labor intensity industries to below-median industries used as concentration measure.
  - Plotting average per capita growth 1980–2000 against 1990 ratio shows a strong positive association.
  - Example averages: emerging Asia and South Asia concentration about 0.6 and higher growth; aid-dependent Africa about 0.47 and lower growth.

### Policy recommendations and mitigation strategies
- Prudent macroeconomic management of aid inflows is essential; specific considerations include:
  - Improve the quality and effectiveness of government spending so productivity improvements offset competitiveness losses.
  - Build supply-side capacity before large-scale aid absorption (e.g., expand skilled labor supply gradually).
  - Consider phased aid delivery—start slow and accelerate as absorptive capacity is built.
  - Consider sterilization options to mitigate real exchange rate appreciation, while weighing costs (e.g., higher domestic interest rates and potential crowding out).
  - Encourage aid-giving nations to reduce impediments to exports from poor countries, and encourage recipient countries to reduce trade barriers; trade can help absorb aid.
  - Adopt experimental approaches to aid delivery with careful attention to delivery mechanisms, incentives, and spillovers.
- Trade-offs:
  - Sterilization and reserve accumulation mitigate real appreciation but may reduce the immediate developmental benefit of spending.
  - Massive short-term expansions in education/training may create short-run adverse effects even if beneficial long-term; hence a paced approach is recommended.

### Summary conclusions
- Evidence supports a credible channel whereby aid inflows cause real exchange rate appreciation and harm labor-intensive and exportable industries via wage and price channels.
- The methodology measures relative sectoral growth effects; net effect of aid on aggregate growth remains an open question and sensitive to specification and horizon.
- Findings raise the bar on the effectiveness required of aid spending and emphasize the need for careful macroeconomic and policy design to avoid adverse competitiveness side effects.

*Source: _wp05126 - Appendix 1 lists all countries for which data were available as well as those countries that*

### Appendix 2. In column (1), outliers are excluded according to the Hadi (1992) procedure.

### _wp05126 - Appendix 2. In column (1), outliers are excluded according to the Hadi (1992) procedure.

### Robustness checks, 1980s (Tables 4A, 5A)
- Table 4A (dependent variable: annual average rate of growth of value added of industry i in country j)
  - Initial industry share (ij): -0.026*** [0.006] (column (1)); -0.025*** [0.005] (col (2)); -0.026*** [0.005] (cols (3)-(5))
  - Aid/GDP (j)*Labor share (i): -0.031** [0.013] (col (1)); -0.026** [0.013] (col (2)); -0.035** [0.015] (col (3)); -0.035** [0.016] (col (4)); -0.035** [0.017] (col (5))
  - Observations: 636 (col (1)); 712 (cols (2)-(5))
  - R-squared: 0.41 (col (1)); 0.39, 0.38, 0.38, 0.38 (cols (2)-(5))
  - Notes: IV estimations; standard errors robust and Frankel and Romer (1999) correction applied; country and industry fixed effects; regressors standardized.

- Table 5A (robustness to measures of aid)
  - Initial industry share (ij): -0.026*** [0.005] (col (1)); -0.026*** [0.005] (col (2)); -0.023*** [0.006] (col (3)); -0.028*** [0.005] (col (4))
  - Aid/GDP(j)* Labor share (i): -0.041** [0.019] (col (1), aid excludes technical assistance); -0.026*** [0.008] (col (2), technical assistance); -0.069** [0.035] (col (3), short impact aid); -0.041*** [0.014] (col (4), long impact aid)
  - Observations: 712 (all columns)
  - R-squared: 0.35, 0.42, 0.21, 0.38 (cols (1)-(4))

### Robustness checks, 1990s (Tables 4B, 5B)
- Table 4B (dependent variable: annual average rate of growth of value added of industry i in country j)
  - Initial industry share (ij): -0.015*** [0.006] (cols (1)-(5) similar)
  - Aid/GDP (j)*Labor share (i): -0.020** [0.010] (col (1)); -0.016* [0.009] (col (2)); -0.020** [0.009] (col (3)); -0.022** [0.008] (col (4)); -0.022* [0.011] (col (5))
  - Observations: 313 (col (1)); 357 (cols (2)-(5))
  - R-squared: 0.34 (col (1)); 0.32 (cols (2)-(5))

- Table 5B (robustness to measures of aid, 1990s)
  - Initial industry share (ij): -0.016*** [0.006] (cols (1)-(3)); -0.015*** [0.006] (col (4))
  - Aid/GDP(j)* Labor share (i): -0.022** [0.010] (col (1)); -0.018** [0.008] (col (2)); -0.025** [0.010] (col (3)); -0.021** [0.009] (col (4))
  - Observations: 357 (all columns)
  - R-squared: 0.32 (all columns)

### Channel: Labor vs Capital (Table 6)
- Table 6 (dependent variable: annual average rate of growth of value added of industry i in country j)
  - 1980s (cols (1)-(2)):
    - Initial industry share (ij): -0.025*** [0.005] (both cols)
    - Aid/GDP(j)* Financial dependence(i): 0.003 [0.010] (col (1)); 0.016 [0.011] (col (2))
    - Aid/GDP(j)* Labor share(i): -0.042*** [0.013] (col (2))
    - Observations: 680 (cols (1)-(2))
    - R-squared: 0.41 (col (1)); 0.38 (col (2))
  - 1990s (cols (3)-(4)):
    - Initial industry share (ij): -0.013** [0.006] (col (3)); -0.016*** [0.006] (col (4))
    - Aid/GDP(j)* Financial dependence(i): -0.01 [0.010] (col (3)); -0.004 [0.011] (col (4))
    - Aid/GDP(j)* Labor share(i): -0.021** [0.009] (col (4))
    - Observations: 343 (cols (3)-(4))
    - R-squared: 0.32 (col (3)); 0.33 (col (4))
  - Note: Financial dependence (i) from Rajan and Zingales (1998).

### Alternative competitiveness indicators (Table 7)
- Table 7 (dependent variable: annual average rate of growth of value added of industry i in country j)
  - Columns (1)-(2): Exportability1 index (i)
    - 1980s: Aid/GDP(j)* Exportability1 index (i): -0.052** [0.023]; Initial industry share (ij): -0.022*** [0.006]; Observations: 712; R-squared: 0.4
    - 1990s: Aid/GDP(j)* Exportability1 index (i): -0.047** [0.020]; Initial industry share (ij): -0.017*** [0.006]; Observations: 357; R-squared: 0.32
  - Columns (3)-(4): Exportability2 index (i)
    - 1980s: Aid/GDP(j)* Exportability2 index (i): -0.077 [0.051]; Initial industry share (ij): -0.025*** [0.005]; Observations: 712; R-squared: 0.38
    - 1990s: Aid/GDP(j)* Exportability2 index (i): -0.038* [0.021]; Initial industry share (ij): -0.014** [0.006]; Observations: 357; R-squared: 0.31
  - Definitions:
    - Exportability1 index (i): dummy = 1 if industry (i) has ratio of exports to value added that exceeds industry median for a group of developing countries.
    - Exportability2 index (i): ratio of exports to value added for industry (i) averaged for a group of developing countries.

### Effects on “Real” value added and employment (Tables 8A, 8B)
- Table 8A (1980s)
  - Columns (1)-(3): annual average rate of growth of value added deflated by sector and country-specific nominal wage rate
    - Initial industry share (ij): -0.023*** [0.005] (col (1)); -0.021*** [0.005] (col (2)); -0.023*** [0.005] (col (3))
    - Aid/GDP(j)* Labor share(i): -0.029** [0.012] (col (1)); -0.021*** [0.008] (col (2))
    - Aid/GDP(j)* Exportability1 index(i): -0.037* [0.021] (employment columns)
    - Aid/GDP(j)* Exportability2 index(i): -0.085* [0.047]; -0.071* [0.040] (relevant columns)
  - Columns (4)-(6): annual average rate of growth of employment of industry i in country j
    - Initial employment(ij): -0.007** [0.003] (col (4)); -0.009*** [0.003] (col (5)); -0.006** [0.002] (col (6))
    - Observations: 633 (all cols)
    - R-squared: 0.32, 0.34, 0.29, 0.29, 0.3, 0.25 (cols (1)-(6) respectively)

- Table 8B (1990s)
  - Columns (1)-(2): value added deflated by sector and country-specific nominal wage rate
    - Initial industry share (ij): -0.019*** [0.006] (col (1)); -0.020*** [0.007] (col (2))
    - Aid/GDP(j)* Labor share(i): -0.016* [0.010] (col (1)); -0.012* [0.007] (col (2))
  - Columns (3)-(4): employment growth
    - Initial employment(ij): -0.009** [0.004] (col (3)); -0.009* [0.005] (col (4)); -0.007* [0.004] (other)
    - Aid/GDP(j)* Exportability1 index(i): -0.053** [0.022] (col (3)); -0.031** [0.015] (col (4))
    - Observations: 303 (cols (1)-(3)); 332 (cols (4)-(6) where applicable)
    - R-squared: 0.24, 0.22, 0.21, 0.27, 0.26, 0.24

### Real exchange rate overvaluation effects (Table 9)
- Table 9 (dependent variable: decade average of annual growth rate of value added of industry i in country j)
  - OLS (1980s): Initial industry share(ij): -0.027*** [0.005]; Overvaluation(j)* Labor Share(i): -0.017*** [0.004]; Observations: 702; R-squared: 0.43
  - IV (1980s): Initial industry share(ij): -0.027*** [0.007]; Overvaluation(j)* Labor Share(i): -0.077** [0.035]; Observations: 702; R-squared: 0.20
  - OLS (1990s): Initial industry share(ij): -0.015*** [0.005]; Overvaluation(j)* Labor Share(i): -0.017*** [0.006]; Observations: 357; R-squared: 0.33
  - IV (1990s): Initial industry share(ij): -0.017*** [0.006]; Overvaluation(j)* Labor Share(i): -0.034** [0.017]; Observations: 357; R-squared: 0.31
  - Note: In columns (2) and (4), instruments for overvaluation are the same as those for the aid variable.

### Impact of remittances (Table 10)
- Table 10 (Impact of Remittances, 1990s; dependent variable: decade average of annual growth rate of value added of industry i in country j)
  - Initial industry share(ij): -0.012** [0.005] (cols (1)-(5) identical)
  - Remittances(j)* Financial dependence(i): 0 [0.007] (col (1)); -0.001 [0.008] (col (2)); 0.001 [0.006] (col (3))
  - Remittances(j)* Labor share(i): 0.004 [0.006] (col (4)); 0.004 [0.007] (col (5))
  - Remittances(j)* Exportability index(i): 0.006 [0.013] (cols (4)-(5))
  - Observations: 343 (all columns)
  - R-squared: 0.32, 0.33, 0.32, 0.33, 0.32 (cols (1)-(5))

### Selected charts and non-parametric results
- Chart 1A (1980s): Core specification conditional correlation
  - coef = -.03191105, (robust) se = .01141765, t = -2.79
- Chart 1B (1990s): Core specification conditional correlation
  - coef = -.01950493, (robust) se = .00873422, t = -2.23
- Chart 2 (Non-parametric depiction of core result, 1980s)
  - coef = -.58726191, se = .24648494, t = -2.38
  - Method: industries split above/below median labor share; for each country, difference in average growth of value added between above- and below-median industries plotted against Aid/GDP(j).
- Chart 3 (Exchange Rate Overvaluation and Aid)
  - Panels for 1970-80, 1980-90, 1990-00 plot Aid to GDP against Real Exchange Rate Overvaluation (definition elaborated in text).
- Chart 4A/4B (Remittances and Exchange Rate Overvaluation, 1990-2000)
  - Chart 4B: remittances for 1990-2000 plotted against real overvaluation for 1980-90.
- Chart 5 (Growth and Concentration of Value-Added in Labor-Intensive Manufacturing)
  - Rate of growth of per capita GDP (1980-2000) plotted against Concentration of value added in labor-intensive sectors (1990); concentration measured as ratio of value-added in sectors above median labor-intensity to those below median.

*Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05126.pdf*

### Appendix 1: Main Data Sources and Description

### _wp05126 - Appendix 1: Main Data Sources and Description

### Main data sources
- UNIDO: Industrial Statistics Database (2003) — data on value added and labor share at 3-digit ISIC (Revision 2). [UNIDO database].
- World Bank: World Development Indicators — data on aggregate manufacturing growth.
- IMF: International Financial Statistics (IFS) — data on remittances and U.S. Producer Production Index used as deflator.
- World Bank: WITS (World Integrated Trade Solution) data — tradability/export data at 3-digit ISIC code.
- OECD: Development Assistance Committee (DAC) — data on aid.

### List of variables and data sources (Appendix Table 1A)
- Growth Rate of Value Added_ij
  - Description: Industry i’s annual growth rate of value added in country j, averaged over each decade.
  - Source: UNIDO (2003).
- Initial Industry Share_ij
  - Description: Industry i’s share in country j’s total manufacturing value added at the beginning of the decade.
  - Source: UNIDO (2003).
- Labor Share_i
  - Description: The labor share index for industry i, measured in terms of the ratio of wage to value added. The industry index was constructed by taking the average across years and countries for each industry and decade.
  - Source: UNIDO (2003).
- Financial Dependence_i
  - Description: The measure of external financial dependence for all firms in industry i during the 1980s.
  - Source: Rajan and Zingales (1998).
- Overvaluation_j
  - Description: The degree of a country’s real exchange rate overvaluation obtained as residuals from a regression of a country’s price level (relative to the United States) on its per capita income level subtracted from the actual price level.
  - Source: Authors’ calculations.
- Exportability1 index_i
  - Description: A dummy that takes a value of 1 if industry i has a ratio of exports to value that exceeds the industry median value. For each industry, the average ratio of exports to value added was calculated using a group of developing countries.
  - Source: WITS data, World Bank (at the 3-digit ISIC code).
- Exportability2 index_i
  - Description: For each industry, it is the average ratio of exports to value added calculated for a group of developing countries.
  - Source: WITS data, World Bank (at the 3-digit ISIC code).
- Aid / GDP_j
  - Description: The ratio of aid to GDP for country j.
  - Source: OECD DAC database.
- Rvamfgpc_j
  - Description: The annual average rate of per capita value added growth in manufacturing.
  - Source: World Development Indicators.
- Remittances_j
  - Description: The ratio of remittances to GDP for country j.
  - Source: IMF’s International Financial Statistics.

### Data construction and measurement details
- Growth Rate of Real Value Added:
  - Nominal value added in US dollars from UNIDO is converted to real value added in constant Year 2000 US dollars using the U.S. Producer Production Index from IMF IFS.
  - The resulting measure is compared with the real value added in local currency to ascertain reliability.
  - Inclusion criterion: correlation between the US-dollar-based real series and the local-currency real series must be higher than 0.75 to be included in the base sample.
  - Where local PPI was unavailable in IFS, alternative deflators were used:
    - Effective deflator constructed with the index of industrial production as in Rajan and Zingales (1998): ratio of growth rate of nominal value added in entire manufacturing sector (UNIDO) to growth rate of index of industrial production (IFS).
    - Alternatively, a GDP deflator was used whenever these two series were not available.
  - Average annual growth rate of real value added for industry i in country j was calculated for the 1980s and 1990s wherever data existed for at least a seven-year period.
- Average Labor Share:
  - For each decade, the average labor share for industry i was computed by taking the unweighted average of the labor share across developing (non-OECD) countries and across years for which data on growth rate of real value added and initial industry share were available in UNIDO.
  - Labor share is defined as the ratio of “wages and salaries” to “value added”.
  - Sample sizes: 954 observations for forty countries are used for the 1980s, and 642 observations for twenty eight countries are used for the 1990s.
- Data cleaning and exclusions:
  - Countries with aid less than 1% of GDP are not included in the regression analysis.
  - Peru is dropped from the base sample due to unusually high level of growth rates in all industries in UNIDO (i.e. exceeds 100% in all sectors).
  - Niger is dropped from the 1990s sample as the data contained an observation where the ratio of wages to value added exceeded 17.
  - Observations where the ratio of wages to value added exceeded one were dropped (resulted in 10 and 12 observations being dropped respectively from the sample for the 1980s and 1990s).
  - Countries whose labor share parameters are not significantly correlated with the average labor share parameter:
    - 1980s: Bolivia, Egypt, Jordan, Mauritius and Zambia.
    - 1990s: Egypt, Ethiopia, and Syrian Arab Republic.

### Aid as a percentage of GDP (Appendix Table 1B) — selected entries
- 1980s (selected top and other entries)
  - Malawi 17.56%
  - Burundi 14.90%
  - Senegal 12.41%
  - Tanzania 11.71%
  - Papua New Guinea 11.09%
  - Zambia 10.72%
  - Madagascar 8.67%
  - Sri Lanka 8.20%
  - Kenya 8.13%
  - Swaziland 7.24%
  - Honduras 6.48%
  - Botswana 6.37%
  - Bolivia 6.15%
  - Bangladesh 5.91%
  - Congo 5.87%
  - Jamaica 5.68%
  - Costa Rica 4.61%
  - Israel 4.03%
  - Mauritius 3.59%
  - Fiji 3.30%
  - Jordan 3.15%
  - Egypt 2.96%
  - Cote d'Ivoire 2.77%
  - Pakistan 2.75%
  - Cameroon 2.59%
  - Ghana 2.38%
  - Tunisia 2.25%
  - Morocco 2.14%
  - Philippines 1.86%
  - Guatemala 1.48%
  - Indonesia 1.17%
  - Thailand 1.01%
  - Malta 0.99%
  - Panama 0.99%
  - Cyprus 0.92%
  - India 0.76%
  - Barbados 0.73%
  - Uruguay 0.37%
  - Algeria 0.33%
  - Chile 0.32%
  - Colombia 0.30%
  - Bahamas 0.19%
  - Singapore 0.16%
  - Korea 0.08%
  - Venezuela 0.07%
  - Iran 0.06%
  - Kuwait 0.03%
- 1990s (selected entries)
  - Tanzania 13.92%
  - Ethiopia 10.83%
  - Senegal 10.40%
  - Bolivia 7.88%
  - Kenya 6.73%
  - Jordan 6.32%
  - Cameroon 4.65%
  - Egypt 3.29%
  - Morocco 1.55%
  - Philippines 1.37%
  - Tunisia 1.32%
  - Mauritius 1.13%
  - Indonesia 1.02%
  - Costa Rica 0.94%
  - Panama 0.81%
  - Uruguay 0.29%
  - Malaysia 0.28%
  - South Africa 0.27%
  - Chile 0.25%
  - Colombia 0.24%
  - Korea 0.04%
  - Singapore 0.03%
  - Venezuela 0.05%
  - Kuwait 0.02%
  - China (Hong Kong) 0.02%
  - India 0.56%
  - Russia 0.46%
  - Cyprus 0.34%
  - Uruguay 0.29%
  - Malaysia 0.28%
  - Oman 0.18%

### Industry labor shares and external financial dependence (Appendix Table excerpt)
- ISIC 311 Food products
  - Average labor share 1980s: 0.36
  - Average labor share 1990s: 0.32
  - External financial dependence: 0.14
  - Additional numeric columns in table: 101.25 and 1.18 (preserved as in source table).
- ISIC 313 Beverages
  - Average labor share 1980s: 0.26
  - Average labor share 1990s: 0.22
  - External financial dependence: 0.08
  - Additional numeric columns: 0.00, 0.14, 0.28.
- ISIC 314 Tobacco
  - Average labor share 1980s: 0.24
  - Average labor share 1990s: 0.18
  - External financial dependence: -0.45
  - Additional numeric columns: 0.00, 0.23, 0.66.
- ISIC 321 Textiles
  - Average labor share 1980s: 0.47
  - Average labor share 1990s: 0.43
  - External financial dependence: 0.40
  - Additional numeric columns: 111.35 and 2.28.
- ISIC 322 Wearing apparel, except footwear
  - Average labor share 1980s: 0.51
  - Average labor share 1990s: 0.46
  - External financial dependence: 0.03
  - Additional numeric columns: 112.46 and 2.21.
- ISIC 323 Leather products
  - Average labor share 1980s: 0.45
  - Average labor share 1990s: 0.44
  - External financial dependence: -0.14
  - Additional numeric columns: 112.65 and 9.57.
- ISIC 324 Footwear, except rubber or plastic
  - Average labor share 1980s: 0.49
  - Average labor share 1990s: 0.48
  - External financial dependence: -0.08
  - Additional numeric columns: 111.69 and 26.81.
- (Table continues for ISIC codes 331 through 390 with average labor shares for 1980s and 1990s and external financial dependence values; entries preserved in the source table.)

### Appendix 2: Instrumentation strategy (summary)
- Identification idea:
  - Aid is often extended for non-economic (strategic) reasons; such strategically-motivated aid is assumed unaffected by recipient economic outcomes.
  - This assumption is used to develop instruments for aid (in the spirit of Frankel and Romer (1999) and exploiting donor-recipient strategic variables).
- Donor allocation model (summarized):
  - Donor d’s share of total aid allocated to recipient r in year t (θ_drt) is modeled as a function of non-economic donor-recipient relationship variables (Y), including:
    - STRAT: 1 if donor and recipient are common members of, or signatories to, an Entente or Alliance in any given time period.
    - USISEG: 1 for US-Egypt and US-Israel observations after the Camp David agreement.
    - COMCOL, COMCOLUK, COMCOLFRA, COMCOLSPA, COMCOLPOR: colonial relationship indicators (ever a colony of the donor; and country-specific colonial dummies).
    - CURCOL: 1 if there is a contemporaneous colonial relationship.
    - COMLANG: 1 if donor and recipient share a common language.
  - Key identifying assumption: none of these right-hand-side variables directly relate to growth in the recipient country.
- Construction of the instrument:
  - Predicted shares (θ̂_drt = Y′β̂) are used to compute predicted bilateral aid flows from donors to recipients:
    - Â_rt = Σ_d (θ̂_drt * A_d_t * GDP_d_t) / GDP_r_t  (expressed as in equation (2) in the source).
  - Â_rt averaged over the relevant period is the instrument used in the paper.
- Exclusion restriction discussion:
  - Concern: instrument variables might have independent impact on growth.
  - Advantage of cross-country, cross-industry framework: country-specific impacts of donor relationships are absorbed in country fixed effects, making it less plausible that donor relationships systematically affect growth in particular industrial sectors.

*Source: Appendix 1 and adjacent appendices of the provided IMF working paper content.*

### References

### References

### Institutional, Growth, and Development
- Acemoglu, Daron, Simon Johnson, James A. Robinson, and Yunyong Thaicharoen (2003) “Institutional Causes, Macroeconomic Symptoms: Volatility, Crises and Growth.” Journal of Monetary Economics 50, 49-123
- Hausmann, Ricardo, Lant Pritchett, and Dani Rodrik. (2004). “Growth Accelerations.” Harvard University.
- Easterly, William, and Ross Levine (2003) "Tropics, Germs, and Crops: How Endowments Influence Economic Development," Journal of Monetary Economics, Elsevier, vol. 50(1), pages 3-39.
- Levchenko, Andrei A. (2004) "Institutional Quality and International Trade," IMF Working Paper No. 04/231
- Rajan, Raghuram and Luigi Zingales (1998) “Financial Dependence and Growth,” American Economic Review, 88 (3): 559-586.
- Rajan, Raghuram and Luigi Zingales, 2003, Saving Capitalism from the Capitalists, Crown Business, New York.

### Foreign Aid, Aid Effectiveness, and Taxation
- Adam, Christopher, and Stephen A. O’Connell (1999) "Aid, Taxation and Development in Sub Saharan Africa", Economics and Politics, vol. 11(3) November: 225-253
- Alesina, Alberto and David Dollar (2000) “Who Gives Foreign Aid to Whom and Why?” Journal of Economic Growth 5 (1): 33-63.
- Alesina, Alberto, and Beatrice Weder (2002), "Do Corrupt Governments Receive Less Foreign Aid?", American Economic Review 92 (4): 1126-1137
- Azam, Jean-Paul, Shantayanan Devarajan, and Stephen A. O'Connell (1999) "Aid Dependence Reconsidered " World Bank Policy Research Working Paper No. 2144.
- Burnside, Craig, and David Dollar (2004) "Aid, Policies, and Growth: Revisiting the Evidence," Policy Research Working Paper Series 3251, The World Bank.
- Clemens Michael A., Steven Radelet and Rikhil Bhavnani (2004) “Counting Chickens when They Hatch: The Short Term Effect of Aid on Growth,” Working Paper No. 44, Center for Global Development.
- Collier, Paul, and David Dollar (2002) "Aid Allocation and Poverty Reduction," European Economic Review, Elsevier, vol. 46(8), pages 1475-1500.
- Dalgaard, Carl-Johan, Henrik Hansen and Finn Tarp (2004). “On the Empirics of Foreign Aid and Growth”, Economic Journal, vol. 114(496): 191-216.
- Easterly, William (2003) “Can Foreign Aid Buy Growth?” Journal of Economic Perspectives 17 (3): 23-48.
- Easterly, William, Ross Levine, and David Roodman (2004) “New Data, New Doubts: A Comment on Burnside and Dollar’s ‘Aid, Policies, and Growth’,” American Economic Review, forthcoming.
- Hansen, Henrik, and Finn Tarp (2001) “Aid and Growth Regressions,” Journal of Development Economics, 64: 547-70
- Prati A., R. Sahay, and T. Tressel (2003), "Is There a Case for Sterilizing Foreign Aid Inflows?", mimeo, July.
- Roodman, David (2004) “The Anarchy of Numbers: Aid, Development, and Cross-country Empirics” Center for Global Development, Washington, DC.
- Svensson, Jakob (2003) “Why Conditional Aid Doesn’t Work and What Can Be Done About It?", Journal of Development Economics, vol. 70 (2): 381-402
- Rajan, Raghuram, and Arvind Subramanian (2005) "Aid and Growth: What Does the Cross-Country Evidence Really Show?" mimeo

### Trade, Production, and Macroeconomic Links
- Frankel, Jeffery, and David Romer (1999) “Does Trade Cause Growth?” American Economic Review, 89 (3): 379-399.
- Nicita, Alessandro, and Marcelo Olarreaga (2001) "Trade and Production, 1976–99," Policy Research Working Paper Series 2701, The World Bank.
- Levchenko, Andrei A. (2004) "Institutional Quality and International Trade," IMF Working Paper No. 04/231
- Reinhart, Carmen, and Kenneth Rogoff (2002) “The Modern History of Exchange Rate Arrangements”, NBER working paper 8963.

### Education, Incentives, and Human Capital
- Banerjee, Abhijit, Shawn Cole, Esther Duflo, and Leigh Linden (2004) “Remedying Education: Evidence from two randomized experiments in India.” Mimeo, MIT
- Kremer, Michael, Edward Miguel, and Rebecca Thornton (2004) "Incentives to Learn" NBER Working Paper No. W10971.
- World Bank (2004) “Making Services Work for Poor People,” World Development Report: Washington D.C.
- World Bank (1998) “Knowledge for Development” World Development Report: Washington D.C.

### Political Economy, Firms, and Organization
- Blanchard, Oliver, and Michael Kremer (1997). “Disorganization,” Quarterly Journal of Economics, 112 (4): 1091-1126.
- Rajan, Raghuram, and Julie Wulf (2004) "The Flattening Firm: Evidence from Panel Data on the Changing Nature of Corporate Hierarchies", mimeo, University of Chicago
- Rajan, Raghuram and Luigi Zingales (1998) “Financial Dependence and Growth,” American Economic Review, 88 (3): 559-586.
- Bauer, P.T. (1971) Dissent on Developmen: Studies and Debates in Development Economics. London. Weidenfeld & Nicholson

### Methodology, Statistics, and Data Sources
- Angrist, Joshua.D., and Alan.B. Krueger, (2001), “Instrumental Variables and the Search for Identification: From Supply and Demand to Natural Experiments,” Journal of Economic Perspectives, Fall, Vol. 15, No. 4, pp. 69-86
- Hadi, Ali, S., (1992), “Identifying Multiple Outliers in Multivariate data,” Journal of the Royal Statistical Society, Series B, 54: 761-771.
- International Monetary Fund (2005) “World Economic Outlook,” Washington; IMF
- United Nations Industrial Development Organization (2003) Database on Industrial Statistics, United Nations, Vienna.

*Content compiled from the PDF chapter titled "_wp05126 - References".*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05126.pdf_