For the social sciences, numbers and figures are at once fascinating and ambiguous. Their validity is often fragile, always open to challenges, but their claim to precision (real or apparent) means that they carry a considerable power of suggestion. Their capacity to order the world thus extends far beyond an academic or scientific framework to form an essential political tool for contemporary societies, as Alain Desrosières has shown. Moreover, their universalization is directly linked to the ever-closer connections established between different parts of the world since the early modern era, followed by the development of international institutions since the end of the Second World War. However, the production of numbers remains an open question, as does their capacity to describe social and economic realities in a reliable way. This is particularly the case when they are applied in a standard fashion to contexts that are profoundly different from those in which they emerged.

These difficulties have returned to the forefront of debate thanks to two incisive and provocative books by Morten Jerven, devoted to the production of quantitative data and the use of figures in research on African economies, whether today or in the distant past. The aim of Jerven's work is to show the weakness of many of the statistics used by development economists. This is particularly the case when it comes to the gross domestic product (GDP) of the anglophone countries of sub-Saharan Africa post-independence (Botswana, Ghana, Kenya, Malawi, Nigeria, Tanzania, Uganda, Zambia), which Jerven studies at length in order to demonstrate the highly questionable principles underlying its fabrication. Calculation of GDP aggregates very diverse kinds of accounting data, and measuring it is a delicate operation that demands a high degree of caution, even in developed countries with high-performing statistical institutes that can draw on a long and stable tradition. The difficulty is even more pronounced in Africa, where economic and social realities are not only very heterogeneous but also juxtapose monetary and non-monetary phenomena to an exceptional extent, and thus elude calculation and quantification by economists. In a number of recent cases, most notably that of Ghana, the GDP of particular African countries has undergone major readjustment after a change in calculation method, leading to a very different evaluation of crucial issues such as the validity of economic policies or the stagnation of African economies. Jerven uses these cases to reflect on the sweeping fragility of quantitative knowledge. In his opinion, the growing attention paid to a global narrative of the African continent leads to the construction of fictions based on unreliable and heterogeneous figures, jeopardizing any possibility of correctly interpreting African economies. He argues for a change in methodology, for a return to the scale of the nation, and for a better historicization, localization, and contextualization of the production of numerical data.

Jerven's work opens up a whole range of questions that are addressed by the articles gathered in this thematic dossier. Of these, three can be usefully highlighted here. The first relates to the sociology of quantification: Who is involved in the production of statistics? Which institutions, public or private, organize, fund, and use the numbers that are generated? This reflection must be very wide in its scope, taking into account not just the (little-studied) national level but also the international scale, since many of the data are produced by foreign experts and organizations (particularly the World Bank) that are not based in the countries whose national accounts they attempt to retrace.

The second question, which lies at the heart of the articles by Agnès Labrousse and Boris Samuel, relates to the uncertainty surrounding the validity of numerical data—a topic on which Jerven's ambiguities are both disconcerting and heuristic. Should we privilege a realist perspective and assume that it is possible to obtain an approximate value for the aggregates, including GDP, that we seek to calculate? If so, this would imply that the difficulty in measuring these figures lies in the mediocrity of African statistical institutions, as well as the administrations of individual countries, incapable of producing and collating statistical data of an economic nature—with the result that international institutions have taken up the baton, themselves producing numerical data whose origins are no less uncertain. From this point of view, our main priority should be to develop the best possible instruments of measurement. Or, on the contrary, should we reflect on what it is we are seeking to measure, and on the equivalences that are drawn between the statistics produced and economic reality? Since the work of Desrosières, who is cited extensively by Labrousse and Samuel, we know that all quantitative measures are conventions, whose contents and contours must be carefully defined if we are to know what we are dealing with. However incomplete it may be, the notion of GDP cannot mean the same thing for countries like France or the United States, where the economy is structured around commercial exchange, as it does for African countries where production and consumption take more diverse forms, and where exchanges between equivalents that can be measured (and taxed by the state) are not always predominant. In other words, it is not simply a question of being able to measure, but also of knowing what it is that is being measured—something that often requires not the expertise of economists but that of, for instance, anthropologists or historians. Vincent Bonnecase has thus drawn our attention to the need to think carefully about the numbers produced by colonial regimes, not just by questioning how they were produced, but also by considering their meaning within the context of their fabrication, in order to avoid attributing them a function or a signification that was absent at the moment they were developed.

The third question, raised in Denis Cogneau's article, relates to whether it is even possible to analyze African economies and their growth, past and present, using these unreliable and poorly constructed quantitative data. This question is particularly important given the recent tendency among economists to eschew monographs (a format allowing a better evaluation of quantitative estimations and their validity) in favor of international studies based on econometric methods that rely extensively on aggregated data. These studies propose comparisons between countries, but at the price of a major blind spot as to the real content of the data used and their commensurability. According to Jerven, longue-durée economic history raises similar problems, since the grand narratives presented by today's economists concerning the maldevelopment of Africa and the multiple historical divergences that supposedly distance the continent from the principles of a good market economy suffer from the same deficiencies: a lack of identification and content for many of the data used, together with inter-country comparisons founded on weak econometric estimates. Nevertheless, none of these issues—bureaucratic fictions, doubts about the data used to construct GDP, the existence of a political economy based on numbers and their instrumentalization for political ends—are exclusively African. Indeed, many of the questions posed in this dossier extend far beyond the African continent.

These discussions of Jerven's work thus find their place at the heart of the project that has driven the Annales since the journal's foundation. Like the dossier that accompanied the publication of Thomas Piketty's Capital in the Twenty-First Century, they enable a confrontation between the methods and tools of economics and those of history, sociology, and anthropology. Footnote ¹ These discussions also provide an opportunity to reinscribe the social and political uses of the figures produced by these tools of measurement within the perspective of a critical epistemology. Through these exchanges, the social sciences can demonstrate their contribution to a collective reflection on public policies, without renouncing their autonomy.