Gender in the World Bank's Poverty Assessments: Six Case Studies from Sub-Saharan Africa

Since the late 1980s, Poverty Assessments (PAs) have emerged as the most important statements by the World Bank about poverty in particular countries. This paper examines, in some depth and from a gender perspective, a set of PAs from four sub-Saharan African countries, focusing particularly on the description and analysis of poverty in rural areas. The paper begins with an overview of the ways in which, and extent to which, women and/or gender issues are present in the PAs. The overall picture is one of inconsistencies, fragmentation and gaps, both between the six examples that are reviewed and within individual country reports. By far the most common way in which women appear in the PAs is in the guise of female-headed households. Where it is not ignored, gender “becomes visibilized” in many different ways and using a very disparate range of gender concepts. Some PAs write about “women”, others use the language of “gender” and yet others have an elaborate set of concepts, including those of the gender division of labour, gender relations, gender discrimination and so on. There is no attempt to systematized the gender analysis. The final policy sections of the PAs are particularly notable for the absence or highly limited discussion of gender issues.

The inconsistencies between the PAs suggest that the country teams had a good deal of autonomy in how they interpreted and prioritized gender issues, and section 3 of the paper looks at the institutional and organizational context in which the PAs were produced. The origins of the PAs in the 1990 World Development Report, the organization and composition of country teams, and the processes by which the PAs were produced are all explored. Although country teams did have a good deal of autonomy in designing their PAs, they were given guidance on poverty measurement analysis and policy in two significant World Bank documents: the Poverty Reduction Handbook (1992a) and Operational Directive on Poverty Reduction (or OD 4.15, 1991). These contain virtually no gender analysis, and a limited range of unsystematic pointers as to where teams might single out women. As a result, different task managers weighted the significance of gender differently and different teams tended to bring their own approaches to gender issues.

In the absence of a clear analytical framework for understanding gender, and detailed guidance of how to produce a gender-sensitive poverty profile, the treatment of gender in the PAs is effectively driven, on the one hand, by a set of epistemological and methodological choices about measuring poverty, and, on the other hand, by the set of prescriptions for reducing poverty which originate in the 1990 World Development Report.

Sections 4 and 5 focus on the empirical evidence collected in the poverty profiles, and ask why gender appears as it does, or indeed why it is “invisibilized” in certain ways. While all PAs employed data collected in surveys using households as units to establish national poverty lines, some also used various other sources, including evidence collected through participatory techniques. Household surveys rarely provide any intra-household data on gender differences—on incomes, for example. The participatory approaches have greater potential to bring out gender issues, but have mainly been used to support findings from national-level surveys. So even where there were gender findings in the initial participatory poverty findings, these have often disappeared in the final report.

Section 6 asks why gender appears as it does (or, again, is “invisibilized”) in the policy analysis of the PAs. In most of the PAs, there are significant gaps as we move from any evidence on gender and poverty that might be present, to the policy analysis. Unlike the country-specific poverty profiles, the policy sections of the PAs are heavily influenced by peer review, and hence by both the 1990 World Development Report model, as well as by subsequent evolving ideas about poverty within the World Bank. Ideas about gender issues and economic growth and poverty have changed over time, with a particular shift from seeing women as vulnerable groups, where they are targeted by social safety nets, to a contemporary concern with the link between female education and growth and poverty.

We conclude that the accumulating evidence in the gender and development literature—namely that men and women experience poverty differently—has had little influence on these six case studies. The PAs lack any substantial appreciation of the issues raised by the study of gender and poverty in Africa over the last two decades. We contrast approaches that treat poverty statically, as an analysis of categories and characteristics, with those adopting a dynamic analysis of poverty, seeing it as the relational processes of impoverishment or accumulation. The link between gender and poverty lies at the level of process, and social and economic relations. For this link to be established, we suggest that poverty must be analysed as relation and process, as must gender. The key conclusion of the paper is that it is impossible to integrate gender into an understanding of poverty unless the reading of evidence, analysis and policy are all based on these relational processes of impoverishment or accumulation.