|
Abstract
This paper uses detailed data on 942 households across
39 communes in Vietnam to compare several quantitative
and qualitative methods to measure poverty and target
the poor. These methods range from expenditure measurement
to proxy-means testing to the traditional classification
of households that takes place at the local level. The
benchmark against which they are compared is the classification
of households by poverty status resulting from wealth-ranking
exercises. These exercises were conducted in 2003, using
a common methodology, by Non-Government Organizations
(NGOs) and local research institutions with a long-standing
trajectory in the field. A random sample of participants
in these exercises was subsequently surveyed, and the
various methods considered in this paper were applied
to it. The traditional classification of households
used in Vietnam yields the targeting outcomes which
are closest to the benchmark, at both local and commune
levels. Proxy-means testing yields results far from
the benchmark at the household level, which is not surprising
given the difficulty to trace the poverty status to
any well-defined set of household characteristics. But
its aggregation at the commune level (a form of small-area
estimation) is correlated to the benchmark. These findings
suggest that a combination of qualitative methods at
local levels and quantitative methods at aggregate levels
could be the most effective way to target the poor.
|
