Easterly and Levine (1997, 1998) tackled the question of "Africa’s growth tragedy" by designing a more detailed empirical model. Their research suggests that the high degree of ethnic diversity and negative spillovers from neighbours explain Africa’s low economic growth. In my own work I have chosen a different approach to re-examine Africa’s growth performance (Hoeffler, 2002). My research is based on a simple and commonly used economic model of growth and focuses on the different statistical methods. My investigation suggests that if the right statistical method is applied, even a fairly simple growth model can account for Africa’s low growth. In other words there is nothing mysterious about Africa – standard models are well suited to explain the region’s poor economic performance.

Robert Solow’s model of economic growth is one of the most famous models in economics (Solow, 1956). The Solow model is based on a simple production function. Output in the economy is produced by using three inputs: capital, labour and technology. Economic growth depends on the initial level of technology, the rate of technological progress, the initial level of income, the rate of investment in capital and the population growth rate.

In the augmented Solow model a further input in the production process is investment in people, or so called human capital (Mankiw, Romer and Weil, 1992). Investment rates, schooling, initial income and population growth rates can be measured and the hypotheses of the model can be tested empirically. As the model predicts, countries with high investment rates in capital and people experience high growth rates, while high population growth rates reduce economic growth.

Once investment and population growth are accounted for, empirical tests show that countries with high initial incomes experienced lower growth rates, i.e., there should be a catching up mechanism at work. Over time poorer countries should thus converge to the income level of initially richer countries.

Using panel data from 1960–1989 for 85 countries, including 22 in sub-Saharan Africa, I examine the Solow model empirically. I use and compare several different methods. The most commonly used estimation technique is the ordinary least squares (OLS) approach. This method does not take either the omitted variable problem or the endogeneity issues into account. Another widely used method is that of fixed effects estimation, which does account for unobserved differences across countries but not for endogeneity. A further set of results is based on the generalized method of moments estimation (GMM), which allows me to take unobserved differences across countries as well as possible endogeneity into account (Arellano and Bond, 1991; Blundell and Bond, 1998).

My comparison of the three methods shows that the significance of the Africa dummy, i.e., that Africa’s performance cannot be explained, depends on the method used. The GMM results, which address the statistical problems discussed above, suggest that even a simple growth model can fully explain Africa’s performance. Africa’s growth has been low because of initial differences, low investment in capital and people, and comparatively high population growth. My comparison of the different statistical methods also shows that the easy to compute estimations (OLS and fixed effects) provide sensible upper and lower bounds for the model estimates. In other words, even though they are biased they can be useful to compute because we know that the true estimates must lie in between these two. This can be useful for researchers who do not have access to the GMM technology or who just want to gain some idea of the approximate size of the estimates.

Conclusion

In my paper (Hoeffler, 2002) I address the question whether Africa’s growth performance can be accounted for in the framework of the augmented Solow model. Using data for a global sample I use and compare different estimation methods. I argue that GMM estimation is my preferred method of estimation. The commonly found result in the literature – that basic growth models are unable to account for Africa’s low growth performance – is only supported when the statistical problems of unobserved country differences and endogeneity are not taken into consideration. Using a GMM estimation that takes these problems into account, I find that the coefficient on the Africa dummy is insignificant. This suggests that the augmented Solow model can fully account for sub-Saharan Africa’s low growth performance.

Collier and Gunning (1999) discuss a typology of possible causes of slow African growth. They conclude that although geographic characteristics adversely affect Africa’s economic performance, poor policies are mainly to blame for Africa’s growth tragedy. It is thus not destiny that determined Africa’s growth performance during the past 30 years, but policy failures. Unless policies are changed to provide the right incentives for an increase in investment and a reduction in population growth, African income growth rates will remain low and the poorest region will be unable to catch up with the rest of the world.