Analysing Chronic Poverty in Rural Cambodia: Evidence from Panel Data

This paper uses four years of panel data on 793 households collected during 2001–11 to measure chronic poverty in rural Cambodia and to identify its key determinants. A household wealth index—a proxy for long-term welfare—constructed by polychoric principal component analysis is used as welfare indicator. Both ordered logistic and multinomial logistic regression models are adopted to identify the causes of chronic and transient poverty by focusing particularly on five explanatory variables: agricultural land and livestock, demography, human capital, social capital and natural resources. To ensure the robustness of our results, two poverty lines are applied: 40th percentile and 60th percentile of the wealth index. The findings indicate that households experiencing chronic poverty account for only 4–10 percent of the total sample, while transient poverty affects 40–52 percent. Among the total poor households, transient poverty is 84–90 percent. Our ordered logistic regression reveals that the composition of household size, the education of the household head, social capital (i.e. connection with three or more people in the community), agricultural land and livestock are likely to be the most important factors that help the chronically poor to move into better off groups. Common property resources seem to have an opposite effect. Multinomial logistic regression results reconfirm that household composition, particularly the number of children aged 7–14 years and females aged 15–64 years, the education of the household head, agricultural land and livestock play an important role in reducing the likelihood of chronic poverty. It appears that education, agricultural land and livestock would also help to reduce transient poverty. Social capital is likely to be strongly correlated with both transient poverty and being never poor.

Available for download

Working Paper