SAGA Proposal: Dynamic Analysis

SAGA
B16 MVR Hall
Ithaca, NY 14853
(607) 255-8931
Fax (607) 255-0178
saga@cornell.edu

SAGA Research Proposal:

1.3.3 Dynamic Analysis

Our focus on notions of vulnerability and on non-monetary dimensions of poverty strongly suggest that our research employ methods that allow us to focus on dynamics of behaviors and outcomes. Climbing out of poverty — or falling into poverty — is inherently a dynamic process. Households that are lucky or more adept begin to accumulate assets that eventually are sufficient to lift them above the poverty threshold permanently. Other less fortunate households suffer shocks with long-term repercussions that send them spiraling downward into greater poverty. Understanding these processes is key to understanding how policies might help the poor to rise out of poverty, and requires data on households' events and circumstances over time.

The most obvious form of such data are longitudinal and panel surveys, which only very recently have become available for Africa. In such surveys households are interviewed at different points in time. Analysis of poverty dynamics using such data for developing countries is new, but rapid progress is being made on methodologies and treatment of specific statistical issues such as measurement error in income or consumption variables and attrition bias (Hoddinott and Baulch 2000; Deaton 1997). Panel data of sufficient length allow researchers to make a crucial distinction between chronic and transitory poverty. The latter appears to be prevalent in developing countries, with households frequently crossing the poverty threshold in one or the other direction. While important, our greater concern is with the determinants of chronic poverty: what keeps poor households in Africa consistently poor, or below the poverty line on average? That is, we are concerned with long-term economic mobility, upward or downward.

Key determinants of long-term changes in poverty status are likely to include accumulation or disaccumulation of assets; policy-induced changes in returns on those assets; and shocks. In principle, these factors are identifiable from household surveys. In addition, initial conditions are likely to be important and can also be measured to varying degrees in surveys. These include levels of human, social, and physical capital, presence of infrastructure, and access to markets, all of which can facilitate potentially risky investments. The role of shocks in determining long-term poverty (as opposed to the more obvious effects on transitory poverty) is not well understood but potentially very important. Transitory income shocks (due, e.g., to weather or policy) may lead to a fall into permanent poverty, through, for example, distress sales of assets; indeed this possibility is essentially what defines economic vulnerability. Positive shocks may have the opposite effect, lifting households above the poverty threshold permanently. Panel data now offer the possibility of investigating these 'irreversibilities' empirically for Africa.

Our other methodological approaches, mixing qualitative and quantitative methods and multidimensional poverty analyses, offer alternative ways of exploring poverty dynamics. Retrospective interviews can elicit detailed information on events that have influenced the respondents income trajectories over a long period. While limited in terms of sample size, these qualitative approaches can explore subtle dynamic processes that large-scale formal household surveys would overlook. A model for this type of work is the research by Scott (2000) on Chile or the 17-year herd histories reconstructed in southern Ethiopia (Lybbert, Barrett, Desta and Coppock 2001). A focus on capabilities and functionings leads to a consideration of dynamics in dimensions other than income. In addition to the broader perspective, considering the dynamics of health, nutritional status, or education avoids many of the measurement problems that plague intertemporal income or expenditure comparisons, especially price deflation and comparability of survey questionnaires.

Previous Section | Next Section

Return to SAGA Research Proposal Table of Contents

width="200" height="33" border="0">