Public Investment and Economic Performance in Highly Indebted Poor Countries: An Empirical Assessment

Abstract

Understanding how public investment affects economic performance in highly indebted low‐income countries is crucial in order to implement effective fiscal policies for adjustment with growth. In this paper we provide an empirical analysis to investigate the relationship between public investment, private investment and output. A dynamic econometric procedure is implemented on a selected group of Highly Indebted Poor Countries (HIPCs). Our results provide empirical support for the crowding‐in hypothesis and a positive relation between public investment and output.

Key Words:

• Crowding‐in
• crowding‐out
• economic growth
• fiscal adjustment
• highly indebted countries
• public investment

JEL Classification:

• H54
• H63
• O11
• O23

Acknowledgements

We would like to thank Pranab K. Bardhan, Riccardo Fiorito, Silvia Marchesi, Malcolm C. Sawyer and an anonymous referee for helpful comments and suggestions. The usual disclaimer applies.

Notes

1. For details see Khan et al. (1990, p. 157).

2. The denomination ‘HIPCs’ refers to the group of developing countries considered as potentially eligible for the debt relief initiative promoted in 1996 by the G8 countries (see International Monetary Found & World Bank, 2001b).

3. The list of the HIPCs is available from International Monetary Found & World Bank (2001b).

4. A more detailed data source description is in Appendix A.

5. See Chirinko (1993) for a survey and a discussion on different strategies for investment determination modelling.

6. See also Pereira (2001a, b).

7. See, for instance, Aschauer (1989a, b).

8. If a regression is estimated with IP as endogenous and IG (together with others) as explanatory variable, it is a priori required that IP does not exert any effect on the right hand variables, and feedbacks are excluded.

9. All the variables are considered in logarithmic form.

10. See Engle & Granger (1987).

11. See Johansen & Juselius (1990).

12. For a comparative study on alternative cointegration approaches we refer to Gonzalo (1994).

13. Respectively Jarque–Bera test, and first and second order Q‐statistics test on the raw residuals and the squared residuals.

14. Also in our case this conclusion may be deduced by observing the variance–covariance matrix of the residuals obtained from the VECM estimation. The correlation is always positive (with the only exception of Malawi) and quite high. Details are available upon request.

15. However this hypothesis will be successively relaxed in order to test the robustness of the results obtained.

16. We underline that the hypothesis of priority of public over private investment is assumed also in other empirical works on the same topic, see for instance Pereira (2001b).

17. Our central scenario includes the two possible orders (IG,IP,GDP) and (IG,GDP,IP), which provide analogous result if only the impulses on IG are considered.

18. Data are available from World Bank (2001).

19. This result is plausible. Indeed, it is possible that the equation of GDP as dependent on IP and IG better fits the data with respect to the other two equations, once we consider that investment decisions are highly influenced by unforeseeable elements. Standard errors are available upon request.

20. The total elasticities, as defined by Pereira (2001b), are obtained dividing the aggregated response values of each variable by the aggregated response values of public investment.