Poor Numbers, the product of Morten Jerven's doctoral research, is a short book with three key messages: African national income statistics are worse than we thought, they can seriously bias the findings of scholars and policymakers (errors do not cancel each other out), and, for these reasons, African statistical capacity building should be a much greater priority in the development community. Although few people may be surprised by these conclusions, and Jerven is not the first to point them out, the book places the issue unequivocally on the agenda and tries to assess the degree to which African development statistics are unreliable.
The first chapter discusses the general techniques of income growth measurement as conducted under the global standard of the United Nations System of National Accounts (SNA). The nontechnical discussion of what GDP estimates are, how they are created, and where serious biases can occur, makes this chapter especially suitable for assignment in undergraduate courses in a wide range of disciplines. The interdisciplinary setup of the discussion on national income accounting also underlines one of the book's main strengths: by reopening the conversation about African GDP data quality it tries to close the gap between the noncritical data users and those who reject them a priori. Jerven maintains that the different ways in which the data produced by national statistical agencies is modified by the three main data providers—the World Bank Group development indicators, the Penn World Tables, and the Maddison database—can not only lead to disparate and even inconclusive results, but also that this problem is more pronounced for Africa than for other world regions. These claims are, amongst others, supported with a table showing the different relative rankings of African countries in these three databases. A country like Guinea for example is ranked as seventh poorest by Maddison, whereas the Penn tables place this country only one spot below the ten richest African countries (pp. 17–19). Is the interregional inconsistency among the customary GDP measures in ordinal rankings more problematic for Africa than for other developing regions as Jerven suggests? A simple exercise of looking at correlation coefficients of relative rankings between the same series for Africa, Asia, and Latin America does not bear this out. For the African series one finds correlation coefficients between 0.88 (Maddison-Penn Tables) and 0.95 (Penn Tables-World Bank), which are arguably not poor correlations in the first place. When comparing these results to those for Asia, where the correlation coefficients range between 0.86 and 0.88, the impression of a distinctly African problem seems somewhat unfounded. The values for Latin America are indeed higher, ranging between 0.92 and 0.95, but the difference remains marginal. A comparison of the African correlation coefficients to those for Western countries, which fall between 0.94 and 0.96, underlines that inconsistencies among the common measures between the developed and developing world are smaller than implied by Jerven's table.
Whether African national income estimates are indeed systematically “poorer” than those for other regions remains debatable, and Jerven's assessment of their overall quality and sources of inaccuracy matters regardless. Chapter 2 takes a closer look at how the creation of African development statistics in the twentieth century was influenced by the priorities of both producers and users of the data. Jerven, who is well-read in the relevant primary and secondary literature on this topic, documents the challenging buildup of African statistical capacity in the course of the colonial period to its subsequent collapse under structural adjustment in the 1980s, when stringent economic and political reforms seriously reduced the ability of and incentives for African states to systematically collect data on their own development (p. 52). Chapter 3 illustrates these historical trends and their implications for data quality with three illuminating case studies. Jerven argues that the large temporal fluctuations in African statistical capacity should be kept in mind by researchers relying on these numbers to study economic change over time. Both economists and African historians will find these chapters insightful.
Chapter 4 focuses on the future: how can we “fix” African GDP data? If the reader is looking for an answer to what the “right” numbers are, or for specific guidelines on how to solve the problem, he/she may find him/herself somewhat disappointed. The policy recommendations offered in this chapter remain somewhat underdeveloped and the book unfortunately does not offer any alternative (newly constructed) calculations that seem more plausible. That said, no topic of study is ever complete in one book, and perhaps these loose ends will prove fertile ground for future research for this young scholar who is not afraid to address important questions and challenge accepted wisdom.
