Paper long abstract:

This paper takes it as given that the most important evidence and method, in terms of influence in the literature discussing economic performance, is GDP growth explained by cross country regressions. The paper first presents a succinct and critical review of the standard regression model frameworks that have been applied in explaining the growth shortfall in Africa.

Most analysis of economic performance accepts the growth evidence at face value and the scholarly debate is focused on different methodological preferences. The research in this paper differs in this respect because it questions the evidence. The general doubt about the quality of growth data in Africa stands in a glaring contrast with the lack of detailed work establishing exactly how poor the evidence is.

In a comparative case study the national accounting methods through history in Botswana, Kenya, Tanzania and Zambia are analysed. These official data are compared to the data available through Penn World Tables, World Development Indicators and the Maddison data. The statistical basis and methodology for the official data is studied with a special view to display the elements of statistical growth, and discerning which parts of the national accounts are economic evidence as opposed to statistical artefacts.