Foreign aid and economic growth in the developing countries - a  (Page 12/21)

GDP is measured in 1995 constant US$, whereas gross domestic savings, FDI and ODA are available in current US$.Therefore the values of these latter three variables has been converted into constant 1995 US$, using a deflator and thencomputed as a fraction of GDP. The ratio of secondary school pupils to population is used as a proxy for human capital accumulation.The trade variable is measured as the percentage of GDP and the population variable is the population growth rate.

Based on the World Bank (1998) statistics that 1991 is the year when the world foreign aid reached its peak(see Figure 1), the data set is divided into three subsets, i.e. the sub-periods of 1975-1991 and 1992-2000 and the overall periodof 1975-2000 respectively.

Since the focus of the study is the impact of foreign aid on economic growth, I will concentrate on thecoefficients of ODA, the regional dummies and their slope dummies. Other variables will also be discussed from time to time.

Chapter V

Empirical results and policy implications

This chapter describes the regression output for each sub-period 1975-1991 and 1992-2000 as well as the overallperiod from 1975 to 2000. Based on the empirical results, the policy implications are discussed correspondingly.

In total, there are 9 regressions for 3 respective periods meaning that there are 3 regressions for everyperiod. Regressions 1, 2 and 3 are for sub-period 1975-1991, regressions 4, 5 and 6 are for sub-period 1992-2000 and regressions7, 8 and 9 are for the overall period 1975-2000. The regression process is repeatedly similar for each period. In the firstregression of each period, i.e. regressions 1, 4 and 7, the dependent variable, per capita GDP growth, has been regressed onthe explanatory variables, namely initial GDP per capita, SAVING, FDI, ODA, SCHOOL, TRADE and POPULATION. Then regional dummies andtheir slope dummies, i.e ODA*EASTASIA, ODA*SOUTHASIA and ODA*SUBSAHARA, are included in the second regression of eachperiod, i.e. regressions 2, 5 and 8 to explore how they affect the relation between aid and growth. Finally dummy for inland countriesand its slope dummy, ODA*INLAND are included in the last regression of each period, i.e. regressions 3, 6 and 9 to investigate how thegeographical features influence the relation between aid and growth. Those slope dummies indicate by how much the slopecoefficient of countries from the above-mentioned regions differs from that of countries in other regions.

In the regression analysis, the White’s Correction of Standard Errors have been used wheneverheteroskedasticity problem arises.

5.1. Sub-period 1975-1991

In regression 2, the Jarque-Bera statistic is 6.45, which is larger than the critical value at the 5% level,rejecting the null hypothesis that the residual follows normal distribution. Consequently, outliers are found in observations 6,14, 18, 34 and 36. Hence, dummies for these outliers are used to detect this non-normality problem. As a result, the Jarque-Berastatistic then falls to 2.27 showing that the residual now follows normal distribution. No evidence of autocorrelation,multicollinearity or wrong functional form was found based on Breusch-Godfrey, correlation and Ramsey RESET testsrespectively.