Econometric analysis of the effect of Ghana’s increasing external debt on foreign direct investment (FDI)

The study explored the relationship between foreign direct investment (FDI) and several covariates including external debt, GDP per capita income, and gross fixed capital formation, for Ghana from 1980 to 2013. It was conducted following Ghana’s bail out by the International Monetary Fund (IMF) to stabilise the Ghanaian currency (the Cedi) as debt levels soared and the Cedi depreciated precipitously against the Dollar. The primary research question that precipitated the study was: by how much will foreign direct investment inflows in Ghana change percentage-point wise per a unit increase in her external debt stock? In the econometric analysis, two OLS regressions were run: the ‘log-level’ model and the ‘log-log’ model. The coefficients on external debt, GDP per capita and gross fixed capital formation (a proxy for infrastructure) were all statistically significant, excluding the GDP deflator, which though had its expected sign, was insignificant at 10% significance level using ‘log-level’ regression. In the ‘log-log’ model, only external debts and gross fixed capital formation were statistically significant at 5% significance level. The R-squared explained 89.98% and 89.64% of the total variation in FDI using the loglevel and log-log models respectively. The semielasticities of log(FDI) with respect to external debts, GDP per capita and gross fixed capital formation were 0.3%, 0.2% and 7.5% respectively, while the elasticities of log(FDI) with respect to external debts and gross fixed capital formation were -1.56 and 1.38 respectively. The magnitude of the proportionate changes in log(FDI) was very large when the percentage increase in the explanatory became very large. To avoid violating the Gauss Markov assumptions of multiple linear regression, the Breusch-Pagan test for heteroscedasticity using ‘log-level’ was employed by regressing residuals-squared on all the independent variable. A correlation test was also run to ensure that there was no perfect correlation between the independent variables. Overall significance of the regression was tested using the F statistics and there was a strong rejection of the null hypothesis that none of the variation in log(FDI) was explained by the independent variables. The study finds that in order for government to attract more FDI, emphasis has to be placed on fiscal discipline, sound economic policies, and infrastructural development. The study will enable government and policy makers to estimate the percentage point fall in FDI per a unit increase in external debts so as to strategise borrowing.Ashesi University Colleg