BERND HARDEWEG1, HERMANN WAIBEL2, SUWANNA PRANEETVATAKUL3, PHUNG DUC TUNG4
1Leibniz Universität Hannover, Institute of Development and Agricultural Economics, Germany
2Leibniz Universität Hannover, Institute of Development and Agricultural Economics, Germany
3Kasetsart University, Department of Agricultural and Resource Economics, Thailand
4Institute of Policy and Strategy for Agriculture and Rural Development, Center for Agricultural Policy, Viet Nam
Previous studies to measure poverty were based on a static concept of poverty and relied on cross-sectional data for income or consumption expenditure over relatively short periods of time. Static poverty is generally expressed by indicators such as the headcount ratio and the poverty gap. Recent research however has established that these indicators are rather imperfect proxies for the well"=being of poor households and insufficient for devising effective strategies of poverty reduction. Instead, the dynamic nature of poverty must be captured by measures with a predictive capacity to recognise the fact that a household not being poor today might yet be vulnerable to falling into poverty as a result of covariate (e.g. natural disasters, financial crises) and idiosyncratic (e.g. death, unemployment) shocks. As vulnerability of a household is a function of (a) occurrence and severity of risks and shocks and (b) its ability to cope through ex"=post and ex"=ante mechanisms (e.g insurance, credit markets, social institutions) an assessment of vulnerability requires panel data of a large sample of households which are either poor or face the risk of being poor in the future. Collecting such information is a challenge given usually insufficient prior information.
This paper reports the sampling procedure for vulnerability assessment used in the DFG research project (DFG-FOR 756), which aims at advancing the vulnerability concept by establishing an empirical database for 4400 rural households in three provinces in Thailand and Viet Nam. A questionnaire capturing sources and use of annual household net income on the one hand and measuring the effect of past shocks as well as perceived risks on the households' current consumption, income and wealth has been developed and intensively tested. A multi-stage sampling procedure was adopted with probability proportional to size in all but the last stage, in which clusters of a fixed size of 10 households per village were sampled.
In order to capture the higher level of heterogeneity in agro-ecological, economic and demographic conditions, additional criteria such as ecological and economic zones and disproportionate sampling were used for Viet Nam.
Initial testing of the data revealed the option to test several models of vulnerability.
Keywords: Household survey, sampling, Thailand, Viet Nam, vulnerability to poverty
Full paper: http://www.tropentag.de/2007/abstracts/full/467.pdf