Where you start is where you end up

The previous sections identified some stylized facts about long-run GDP per capita growth and the distribution of income at the country level in Africa. These stylized facts - low and volatile growth, the formation of clubs, lack of convergence in income levels and rising inter-country income inequality - point towards little dynamism in income growth. In this section, we examine the income dynamics of countries in greater detail.

We begin by asking the question: How stable is the per capita income of a typical country in Africa? To test for income stability we ran the following regression:

where Y_i is the mean GDP per capita of country i and Y⁷ is the GDP per capita of country i in 1975. Given our prior evidence of little income dynamism, we expect that average income in a typical African economy will be close to its initial income in 1975.¹⁵ If so, the coefficient estimate of в should be close to 1. A region with a large number of dynamic economies would have an estimated coefficient of significantly more than 1, indicating that initial income underpredicts average income.

The result, presented in Figure 8.11, shows a line near 45 degrees (в = 0.901, t = 7.41). Apart from a few cases, the mean GDP per capita 1975-2005 closely mirrors that of 1975, reflecting a high degree of inertia.¹⁶ There are some positive outliers, such as Botswana, Cape Verde, Lesotho, Mauritius and Namibia, whose mean GDP per capita is well above their 1975 levels. Negative outliers such as the Cote d’Ivoire, Democratic Republic of the Congo, Eritrea, Gabon,

Figure 8.11 Mean gross domestic product (GDP) per capita as a function of GDP per capita in 1975

Source: Authors’ computations.

Madagascar and Zambia, had mean GDP per capita well below their initial levels of income.¹⁷

To check whether there was a change in income stability accompanying the growth acceleration after 1995, we ran the same stability tests for mean GDP per capita for the subperiods 1975-1994 and 1995-2005, by regressing average income in those periods on the initial level of income in 1975. The results are highly significant and close to 1 (estimated coefficients of 0.891 and 0.909, respectively). These are virtually identical to the results of Equation 8.5, implying that the income structure remained highly stable, even after the break in the income and growth series in around 1995.