## 1.     ECCU: Rate of Return on Public Investment

## Source details

**Canonical URL:** [1.     ECCU: Rate of Return on Public Investment](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp07124.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp07124.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp07124.pdf.json)

---

### Introduction and context
- Public investment in the Eastern Caribbean Currency Union (ECCU) averages around 10 percent of GDP, compared to 3 percent for OECD countries.
- Sample period: 1975–2005 (annual data).
- ECCU in this paper: the six independent states of Antigua and Barbuda, Dominica, Grenada, St. Kitts and Nevis, St. Lucia, and St. Vincent and the Grenadines.
- Model purpose: quantify the effect of public investment on growth and the bilateral real exchange rate (RER) with the United States.
- Major endogenous variables: real GDP, real public investment, bilateral RER (with the United States).
- Exogenous controls: OECD growth (contemporaneous and lags), aid flows (explicit aid, deflated by the investment deflator), dummy variables for natural disasters and elections.

### Estimation approach and identification
- Model: panel VAR (PVAR) estimated in first differences of growth rates with fixed effects swept away via the Holtz-Eakin, Newey, and Rosen (1988) approach.
- Instrument set (z): [x_{t-1}, x_{t-2}, w_{t}, w_{t-1}, w_{t-2}, f_t].
- Estimation method: Generalized Method of Moments (GMM) with a weighting matrix robust to unknown-form heteroscedasticity.
- Lag length: two lags (based on log-likelihood tests and information criteria).
- Ordering for orthogonalized impulse responses: public investment → GDP → RER (public investment leads GDP; GDP leads RER).
- Impulse response standard errors derived by Monte Carlo simulations (Hamilton, p.337, 1994).

### Main empirical findings — effects of public investment
- Short-run impact:
  - Public investment has a positive but short-run impact on growth.
  - A one-time random one-standard deviation shock on the growth rate effectively dies out after 4 years.
  - Major effect is direct (boosting domestic demand) with relatively weak private-sector multipliers.
- Long-run (rate-of-return) estimates (Pereira (2000) metric adapted; 20-year horizon with constant output flow):
  - Baseline model:
    - Long-run Elasticity (upper bound): 0.08
    - Marginal Productivity: 0.76
    - Annual Rate of Return: -2.4
    - Long-run Elasticity (central tendency): 0.07
    - Marginal Productivity: 0.66
    - Annual Rate of Return: -3.6
    - Long-run Elasticity (lower bound): 0.06
    - Marginal Productivity: 0.54
    - Annual Rate of Return: -5.3
  - Including aid flows as an exogenous regressor:
    - Central tendency Long-run Elasticity: 0.05
    - Marginal Productivity: 0.47
    - Annual Rate of Return: -6.3
  - Including aid flows and a time trend:
    - Central tendency Long-run Elasticity: 0.07
    - Marginal Productivity: 0.61
    - Annual Rate of Return: -4.3
- Comparison with literature (20-year "bullet" payoff assumption):
  - ECCU: Rate of Return = -2.0
  - U.S. (Pereira (2000)): 7.8
  - Sample of euro-area countries (Pereira and Fatima-Pinho (2006)):
    - Average (excl. ECCU): 4.5
    - Standard deviation (excl. ECCU): 6.2

### Causality and dynamic responses
- Granger causality:
  - For all three model specifications, the hypothesis that public investment did not Granger-cause output growth is rejected at the 1 percent level of significance.
- Growth → public investment:
  - A growth shock has a positive effect on public investment; standard errors are wide.
- Bilateral RER responses:
  - The bilateral RER appreciates in the first three years following positive public investment and growth shocks; standard errors are very wide.
  - In the long-term, higher levels of public investment appear to have little effect in reversing the short-term appreciation.
- Natural disasters and elections (estimated coefficients, Table 3):
  - Natural disaster (same year)
    - Public Investment coefficient: -36.4; standard error: 32.5
    - Real GDP coefficient: -6.2; standard error: 2.5
    - Bilateral RER coefficient: 2.9; standard error: 1.9
  - Natural disaster (following year)
    - Public Investment coefficient: 24.6; standard error: 36.8
    - Real GDP coefficient: 16.5; standard error: 6.8
    - Bilateral RER coefficient: -1.7; standard error: 2.4
  - Election (preceding year)
    - Public Investment coefficient: 12.1; standard error: 15.0
    - Real GDP coefficient: 1.2; standard error: 1.2
    - Bilateral RER coefficient: 0.3; standard error: 0.8
  - Election (same year)
    - Public Investment coefficient: 20.0; standard error: 19.4
    - Real GDP coefficient: 0.6; standard error: 1.4
    - Bilateral RER coefficient: -0.8; standard error: 0.8
- Growth dynamics around disasters:
  - Growth falls sharply in the year of a disaster and rises strongly the following year (consistent with reconstruction dynamics).

### Data and descriptive statistics
- Public investment definition: capital expenditure by central government, public enterprises, and statutory corporations.
- Public investment deflator: average OECD deflator of capital expenditure for the federal government (excluding defence), and state and local governments.
- Public investment ratio volatility:
  - ECCU average public investment-to-GDP ratio standard deviation: 3 percent.
  - OECD average standard deviation: 0.3 percent.
- Stationarity and cointegration:
  - Both the log of real GDP and the bilateral RER are non-stationary.
  - Evidence that public investment is I(1) is weaker and may be driven by one country in the panel.
  - Pedroni (1999) panel cointegration tests were mixed; in the most general specifications there was little evidence of panel cointegration.
- Data sources: IMF staff reports and World Bank Country Economic Memorandums (World Bank source important for pre-1995 PSIP information).

### Conclusions and policy-relevant implications
- Public investment in the ECCU played an important role in raising and sustaining economic activity, but observed effects on growth over the last 30 years have been temporary and limited.
- The estimated rate of return on public investment for the ECCU is likely negative under the metrics used, implying that each EC dollar of public investment induced less than one EC dollar of cumulative real output change.
- Fiscal implications:
  - If public investment is financed by borrowing, it may have contributed more to the debt stock than to GDP growth.
- Recommendations for future research and policy focus:
  - Assess why public investment appears relatively inefficient in the ECCU.
  - Differentiate the effects of various types of public investment (infrastructure vs. social services) on growth and competitiveness.
  - Account for natural-disaster-related depreciation/destruction rates when estimating public capital stocks and returns.

*Source: _wp07124 - References, author calculations and reported tables and figures.*

### 1.     ECCU: Rate of Return on Public Investment .................................................................9

### 1.     ECCU: Rate of Return on Public Investment .................................................................9

### Major sections
- 1.     ECCU: Rate of Return on Public Investment .................................................................9
- 2.     ECCU: Rate of Return Compared.................................................................................10

### Appendices
- I.      ECCU      Public      Investment      Model: Summary Statistics .................................................12
- II.    Correlation    Matrix    of Model Variables ........................................................................12
- III.   VAR Coefficients on Natural Disaster and Election Dummy Variables ......................12

### Figures
- 1.     ECCU: Public Investment, Growth and Aid Inflows, 1975–2004................................13
- 2.     ECCU: Public Sector Investment, 1975–2004..............................................................14
- 3.     ECCU: Impulse Response for GDP and the Bilateral Real Exchange Rate to One 
  Standard Deviation Public Investment and GDP Shocks .............................................15
- 4.     ECCU: Accumulated Impulse Responses to a One Standard Deviation Shock to Public 
  Investment ...............................................................................................................16
- 5.     Accumulated Impulse Responses to a One Standard Deviation Shock to GDP...........17
- 6.      Accumulated Impulse Responses of the Bilateral RER to a One Standard Deviation 
  Shock to GDP and Public Investment .....................................................................18

*Source: _wp07124 - 1.     ECCU: Rate of Return on Public Investment .................................................................9*

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

### _wp07124 - References

### Introduction and context
- Public investment in the Eastern Caribbean Currency Union (ECCU) averages around 10 percent of GDP, compared to 3 percent for OECD countries.
- The study uses annual data from 1975 to 2005.
- The ECCU in this paper refers to the six independent states of Antigua and Barbuda, Dominica, Grenada, St. Kitts and Nevis, St. Lucia, and St. Vincent and the Grenadines.
- The paper employs an atheoretic panel vector autoregression (PVAR) to quantify the effect of public investment on growth and the bilateral real exchange rate (RER) with the United States.
- Major endogenous variables: real GDP, real public investment, bilateral RER (with the United States).
- Exogenous controls: OECD growth (contemporaneous and lags), aid flows (explicit aid, deflated by the investment deflator), dummy variables for natural disasters and elections.

### Estimation approach and identification
- Model: panel VAR estimated in first differences of growth rates with fixed effects swept away via the Holtz-Eakin, Newey, and Rosen (1988) approach.
- Instrument set (z) uses lags of endogenous variables and exogenous controls: [x_{t-1}, x_{t-2}, w_{t}, w_{t-1}, w_{t-2}, f_t].
- Estimation method: Generalized Method of Moments (GMM) with a weighting matrix robust to unknown-form heteroscedasticity.
- Lag length chosen: two lags (based on log-likelihood tests and information criteria).
- Orthogonalized impulse responses ordered: public investment → GDP → RER (public investment leads GDP; GDP leads RER).
- Impulse response standard errors derived by Monte Carlo simulations (Hamilton, p.337, 1994).

### Main empirical findings — effects of public investment
- Short-run impact:
  - Impulse responses indicate public investment has a positive but short-run impact on growth.
  - The effect of a one-time random one-standard deviation shock on the growth rate effectively dies out after 4 years.
  - Evidence suggests the major effect of public investment on growth is direct (boosting domestic demand) with relatively weak private-sector multipliers.
- Long-run (rate-of-return) estimates (Pereira (2000) metric adapted; results based on a 20-year horizon with constant output flow):
  - Baseline model
    - Long-run Elasticity (upper bound): 0.08
    - Marginal Productivity: 0.76
    - Annual Rate of Return: -2.4
    - Long-run Elasticity (central tendency): 0.07
    - Marginal Productivity: 0.66
    - Annual Rate of Return: -3.6
    - Long-run Elasticity (lower bound): 0.06
    - Marginal Productivity: 0.54
    - Annual Rate of Return: -5.3
  - Including aid flows as an exogenous regressor
    - Central tendency Long-run Elasticity: 0.05
    - Marginal Productivity: 0.47
    - Annual Rate of Return: -6.3
  - Including aid flows and a time trend
    - Central tendency Long-run Elasticity: 0.07
    - Marginal Productivity: 0.61
    - Annual Rate of Return: -4.3
- Comparison with literature (20-year "bullet" payoff assumption):
  - ECCU: Rate of Return = -2.0
  - U.S. (Pereira (2000)): 7.8
  - Sample of euro-area countries reported in Pereira and Fatima-Pinho (2006) produce a range and an Average (excl. ECCU) of 4.5 and Standard deviation (excl. ECCU) of 6.2.

### Causality and dynamic responses
- Granger causality: For all three model specifications, the hypothesis that public investment did not Granger-cause output growth is rejected at the 1 percent level of significance.
- Growth → public investment: A growth shock has a positive effect on public investment, consistent with higher growth supporting tax and other revenues and enabling increased capital expenditure; standard errors are wide.
- RER responses:
  - The bilateral RER appreciates in the first three years following positive public investment and growth shocks; standard errors are very wide.
  - In the long-term, higher levels of public investment appear to have little effect in reversing the short-term appreciation, suggesting short-run demand effects on prices dominate supply effects.
- Natural disasters and elections:
  - Natural disaster dummies (identified when a disaster affected at least 25 percent of the population or caused more than 25 percent of GDP in damage) were included for the year of the disaster and the two following years to capture reconstruction effects.
  - Estimated coefficients (Table 3):
    - Natural disaster (same year)
      - Public Investment coefficient: -36.4; standard error: 32.5
      - Real GDP coefficient: -6.2; standard error: 2.5
      - Bilateral RER coefficient: 2.9; standard error: 1.9
    - Natural disaster (following year)
      - Public Investment coefficient: 24.6; standard error: 36.8
      - Real GDP coefficient: 16.5; standard error: 6.8
      - Bilateral RER coefficient: -1.7; standard error: 2.4
    - Election (preceding year)
      - Public Investment coefficient: 12.1; standard error: 15.0
      - Real GDP coefficient: 1.2; standard error: 1.2
      - Bilateral RER coefficient: 0.3; standard error: 0.8
    - Election (same year)
      - Public Investment coefficient: 20.0; standard error: 19.4
      - Real GDP coefficient: 0.6; standard error: 1.4
      - Bilateral RER coefficient: -0.8; standard error: 0.8
  - Growth falls sharply in the year of a disaster and rises strongly the following year (consistent with reconstruction dynamics).

### Data and descriptive statistics
- Sample period: 1975–2005 (annual data).
- Public investment definition: capital expenditure by central government, public enterprises, and statutory corporations.
- Data sources: IMF staff reports and World Bank Country Economic Memorandums (World Bank source important for pre-1995 PSIP information).
- Public investment deflator: average OECD deflator of capital expenditure for the federal government (excluding defence), and state and local governments.
- Public investment ratio volatility:
  - ECCU average public investment-to-GDP ratio shows a standard deviation of 3 percent.
  - OECD average standard deviation: 0.3 percent.
- Additional summary points:
  - Both the log of real GDP and the bilateral RER are non-stationary.
  - Evidence that public investment is I(1) is weaker and may be driven by one country in the panel.
  - Pedroni (1999) panel cointegration tests were mixed; in the most general specifications there was little evidence of panel cointegration.

### Conclusions and policy-relevant implications
- Public investment in the ECCU has an important role in raising and sustaining economic activity, but its observed effects on growth over the last 30 years have been temporary and limited.
- The estimated rate of return on public investment for the ECCU is likely negative under the metrics used, implying that each EC dollar of public investment induced less than one EC dollar of cumulative real output change.
- Implication for fiscal policy: If public investment is financed by borrowing, it may have contributed more to the debt stock than to GDP growth.
- Recommendation for future research and policy focus:
  - Assess why public investment appears relatively inefficient in the ECCU.
  - Differentiate the effects of various types of public investment (infrastructure vs. social services) on growth and competitiveness.
  - Account for natural-disaster-related depreciation/destruction rates when estimating public capital stocks and returns.

*Source: _wp07124 - References, author calculations and reported tables and figures.*

### REFERENCES

### REFERENCES

### Public capital, infrastructure, and growth
- Aschauer, D.,1989a, “Is Public Expenditure Productive,?” Journal of Monetary Economics, 23, pp. 177–200.  
- Aschauer, D., 1989b, “Does Public Capital Crowd Out Private Capital,” Journal of Monetary Economics, 24, pp. 171–88.  
- Jorgenson, D., 1990, “Fragile Statistical Foundations,” The Public’s Capital, pp. 6–7.  
- Kamps, Christophe, 2004, The Dynamic Macroeconomic Effects of Public Capital—Theory and Evidence for OECD Countries, Springer-Verlag, Berlin.  
- Kavanagh, Catherine, 1997, “Public Capital and Private Sector Productivity in Ireland, 1958–90,” Journal of Economic Studies, Vol. 24, No. l/2, pp. 72–94.  
- Khan, Mohsin S., and Manmohan S. Kumar, 1997, “Public and Private Investment and the Growth Process in Developing Countries,” Oxford Bulletin of Economics and Statistics, Vol. 59, No. 1, pp. 69–88.  
- Munnell, A., 1992, “Infrastructure Investment and Economic Growth,” Journal of Economic Perspectives, 6, pp. 189–98.  
- Pereira, Alfredo M., 2000, “Is All Public Capital Created Equal?,” Review of Economics and Statistics, Vol. 82, No. 3, pp. 513–18.  
- Pereira, Alfredo M., and Maria de Fatima-Pinho, 2006, “Public Investment, Economic Performance, and Budgetary Consolidation: VAR Evidence for the 12 Euro Countries,” College of William & Mary, Working Paper No. 40.  

### Regional and policy applications (Caribbean, Windward Islands, ECCU)
- Duttagupta, Rupa, and Guillermo Tolosa, 2006, “Fiscal Discipline and Exchange Rate Regimes: Evidence from the Caribbean,” IMF Working Paper, 06/119, (Washington D.C.: International Monetary Fund).  
- Mlachila, Montfort, and Paul Cashin, 2007, “The Macroeconomic Impact of Trade Preference Erosion on the Windward Islands,” Eastern Caribbean Currency Union: Selected Issues, IMF Country Report No. 07/97, pp. 18–43 (Washington D.C.: International Monetary Fund).  
- Roache, Shaun K., 2006, “Domestic Investment and the Cost of Capital in the Caribbean,” IMF Working Paper 06/152 (Washington D.C.: International Monetary Fund).  

### Econometric methods and dynamic panel analysis
- Hamilton, James, D., 1994, “Time Series Analysis,” Princeton University Press, Princeton, New Jersey.  
- Holtz-Eakin, D., Newey, W., and H. S. Rosen, 1988, “Estimating Vector Autoregressions with Panel Data,” Econometrica, 56, pp. 1371–95.  
- Judson, R., and A. L. Owen, 1996, “Estimating Dynamic Panel Data Models: A Practical Guide for Macroeconomists,” Federal Reserve Board of Governors, Washington D.C.  
- Nickell, Stephen, 1981, “Biases in Dynamic Models with Fixed Effects,” Econometrica, 49, pp. 1417–26.  

*Source: _wp07124 - REFERENCES*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp07124.pdf_