Using asset poverty measures to understand poverty dynamics, poverty traps and farmer behavior in sub-Saharan Africa: A focus on rural Ethiopia.

摘要

Effective poverty reduction programs require careful measurement of poverty status. Commonly used consumption or income-based classifications of poverty aggregate together households that are persistently poor with those who are only in poverty due to passing conditions. They also classify as non-poor households that are at risk of falling into poverty as well as those that are not at risk. The tendency to group households that are likely to exit poverty independently with other poor households who lack this ability undermines the targeting of interventions to alleviate poverty and distorts evaluation of anti-poverty programs. Asset-based poverty measures enable more nuanced identification of poverty status, but these methods raise methodological problems when estimating the relationship between assets and livelihood. This dissertation uses panel data from Ethiopia to generate an asset-based poverty classification scheme. Regression results are used to derive an asset index and classify households into various categories of poverty. Asset index dynamics are also explored to test for the existence of multiple asset index equilibria evidence of poverty traps. Results provide evidence of multiple equilibria in the study sample as a whole as well as convergence at different levels for different peasant associations, depending on commercialization opportunities and agro-ecological factors. The asset-based poverty classifications predict future poverty status more accurately than income-based measures implying that the asset-based measure could be used to more carefully target poverty interventions and to more accurately assess the impact of those interventions.Microfinance is often touted as a practical means of helping rural poor overcome capital constraints, and invest in new technology. Using an asset-based approach to poverty measurement and classification chapter three of this dissertation asks whether microfinance has a differential impact on use of improved technology and on consumption and asset growth depending on the family's asset poverty status. The analysis finds no relationship between participation in microfinance programs and the use of modern technologies for the poorest households. Microfinance has a positive direct effect on both consumption and asset growth as well as on the use of modern technology among the relatively wealthier (less poor) households. I find that households who use fertilizer tend to enjoy more rapid consumption growth, and greater accumulation of productive assets, irrespective of their poverty status, but microfinance has no effect on the likelihood of fertilizer use among the poorest households. This implies that while modern technology could present a pathway out of persistent poverty, current formal credit programs are not serving the poorest households in this endeavor. The findings confirm the need to closely assess constraints faced by different classes of poor households and suggest the value of asset based poverty classifications in identifying target groups.The adoption and use of modern technologies is generally accepted as a potential vehicle out of poverty but adoption rates in Ethiopia remain low with the nature of the adoption process largely unstudied (Spielman, 2007). Chapter 4 of this dissertation studies the impact of social networks and social learning on technology adoption in rural Ethiopia. Considering the potentially different marginal benefits of reducing information constraints by poverty status and technology type, the chapter explores the differential impacts of social networks by network type, technology and the asset poverty status of households In addition to geographic networks, it considers the role played by networks with more purposeful interactions such as a household's friends. Results confirm the presence of social learning among farmers in rural Ethiopia, with significant difference across network type, farmer type and technologies. Social learning occurs in networks with purposeful interaction and depending on the technology this effect differs across households experiencing different degrees of poverty.