In his book Poor Numbers: How We are Misled by African Development Statistics and What to Do About It, Morten Jerven has attempted to address an important question regarding the quality and reliability of the numbers which are used to inform development statistics with respect to Sub-Saharan Africa (SSA). According to the author, the short answer is that the numbers used to describe and define African economies are ‘poor’.

Poor Numbers should be a welcome addition to the literature in this area, and not just relevant to the SSA region. Social scientists and policymakers all over the world are too eager to build economic models and development policies without taking into account the quality of the data that underpins them, taking it for granted that the data provided represent real facts. This book strengthens our understanding of this existing problem and directs our attention towards issues of data quality that are not unique only to the SSA region.

The first chapter reveals core issues of data quality such as arbitrary and murky decisions about baseline estimates and it also examines problems within the process of aggregation. It introduces a comparison of the ranking of African countries by income per capita which reflects well the fragility of our statistical knowledge. Jerven argues that ‘Angola … Congo-Brazzaville, Nigeria and Zambia make leaps of more than ten places from one source to the other, leaving the relative ranking of one-fifth of the countries as a matter of high uncertainty’ (p. 19). Given that there are so many different methods and adjustments used to generate these statistics ‘any upward or downward adjustment can find a technocratic … justification’ (p. 29).

The second chapter states that income statistics need to be fully historicised and contextualised. Touching on the period of structural adjustment in the 1980s and 1990s, this chapter criticises the growth-oriented reforms of the IMF and the World Bank. Jerven mirrors the observation of Ward who – along with many others – has strongly criticised the structural adjustment reform and passed judgements on ‘the adoption of these more fragile but ‘up to date’ figures prepared by the Bank’ (Ward Reference Ward2004: 100). The thesis of this chapter is straightforward; data are not facts but products which are highly subject to particular economic and political constraints.

The aim of Chapter 3 is to show that a political economy in which ‘facts’ are embedded is crucial. An indicative selection of case studies makes the link more explicit between statistics and the political economy as it illustrates well ‘how malleable the data are and how political conditions … affect the data’ (p. 56).

Drawing on all of these insights, Chapter 4 emphasises that the problem with development statistics is both a knowledge problem and a governance problem. Jerven argues that all of these findings lead to the need for ‘meta-data’ which ensures a certain degree of transparency about how data are collected, how adjustments are made and how the re-basing of indices are carried out. Data users should be enabled to judge whether a large fluctuation reflects economic change or a statistical error and should be able to acquire some understanding of what extent the dataset corresponds with what is known about a country. Chapter 4 further shows that there is no coherent global strategy for improving the provision of data and that the development community should turn its attention to the important role of local statistical offices.

This publication has several laudable features. First, the author's use of language throughout the argument is very accessible and easy to follow for a non-expert in development statistics. Whilst it does not provide a rigorous treatment of statistics it makes a vibrant and strong argument. Second, the author uses a multidisciplinary framework, combining the analytical tools of the economic historian and the work of economic anthropologists and political scientists, which is crucial if one wishes to understand the multifaceted nature of statistics. Furthermore, Jerven's fieldwork is remarkable. His visits to local statistical offices and his meetings with representatives of the World Bank and the IMF elucidate the issue from a unique standpoint. This book is very original in the sense that all findings, interviews and field visits are largely the result of the efforts of a single individual.

These are all very valuable contributions, but the book also has some weak points. Chapter 1 lists all the important issues which is a good start, yet very little attention has been paid to conceptual accuracy. Jerven uses interchangeably the terms economic development, economic activity, and economic performance, and uses GDP and wealth as equivalent concepts. This loosely employed terminology is an essential shortcoming, given that the book articulates many times that statistics need conceptual and methodological rigour and precision. Furthermore, Jerven should have taken into account and analysed more contributory factors, for instance the impact of the institutions or how the changing assumptions of statistical institutions affect what and how we measure. Certain factors such as the ever-changing nature of political priorities, the end of a political hegemony and geopolitical reorganisation have all influenced the way that we measure and define growth and these too should have been considered. Another weak spot is that Chapter 4 introduces the meta-data as a cure-all for the listed issues. Even if it is a promising idea to use meta-data, it is surprising to some extent that Jerven, who expresses his scepticism regarding the reliability of data, is willing to accept the validity and accuracy of information attached to meta-data. Furthermore, most of the analysis in the book is based on research in Anglophone countries. The author makes conclusions in respect of the whole Sub-Saharan African region, whilst some countries were left out of the analysis.

In summary, Poor Numbers is a pioneering publication, the main message of which is that, even if social scientists, policymakers and other data users never understand completely how and under what circumstances the numbers they are using are generated, at least they need to be aware of the possible inherent errors and to have a critical approach to using numbers in their analyses and in the policies they advocate.