According to several studies, poverty in rural Ethiopia has fallen significantly since the early 1990s, thanks to improved governance and economic liberalisation policies. This paper presents several arguments that challenge this view. The first questions the methodological foundations of the data from which these positive trends are derived: we argue that the original sampling frame was too small and unrepresentative to provide a basis for extrapolating national poverty levels or trends. The second argument questions the conceptual underpinnings of these studies: poverty estimates based on levels of current consumption fail to allow for non-income dimensions of wellbeing, nor for confounding factors such as seasonality, annual rainfall and food aid receipts. The third strand considers alternative sources of data on changes in wellbeing in Ethiopia: recent qualitative studies report that the poor perceive themselves as poorer and more vulnerable than poverty headcount figures suggest.
Finally, we report findings from our own survey in chronically poor and historically famine-prone Wollo. First, a significant proportion of households in the study area are destitute - destitution being defined as inability to meet basic needs, lack of key productive assets, and dependence on transfers. Secondly, the numbers of destitute people, and of people vulnerable to becoming destitute, have increased over the past ten years. Thirdly, the crisis of livelihoods underlying this trend is affecting entire communities - the dominant pattern is an aggregate downward shift, rather than stratification - and the decline of wealthier households is exacerbating the vulnerability of the poorest. These findings cast serious doubts on generalisations about poverty trends in Ethiopia. At the very least, national-level data need to be disaggregated: improving national trends may conceal pockets of entrenched poverty and a deepening livelihoods crisis in parts of rural Ethiopia.