Page 118 - Contributed Paper Session (CPS) - Volume 7

CPS2043 Adnan Dawood K. B. et al.
capita expenditure to be used in determining whether the household is poor
or not poor. As known, the results of the PMTF is overestimated for the poorest
segment of households, whom their average per capita expenditure is very
low. Because the estimation is a linear, it will always be above the lowest
households expenditure, whom are our target.
When building the PMTF for targeting the poor households, it is used to
use part of the population in order to enhance the efficiency of the formula.
The households whom their expenditure is high will be ignored during the
process of building the formula. The most common approach in constructing
the PMTF for targeting the poor households is to play with the covered size of
the population during the running process of the linear regression. Some of
researchers include 50 percent of the population; some includes 30 percent of
the population and so on. In fact, poverty incidence and poverty line play a
key role in determining the size of population that should be covered in the
running process. An accurate linear regression should be derived to predict
the annual per capita expenditure, especially when we are targeting the
poorest of the poor (the poorest of the abject poor).
As known, there are two specific indicators playing a major role in
accepting or rejecting the efficiency of the formula. The first indicator is the
leakage rate and is defined as: the proportion of households who are in reality
non-poor and became poor by the formula (errors of inclusion), the second
indicator is the under-coverage rate and is defined as: the proportion of
households who are in reality poor and became not poor by the formula
(errors of exclusion). The predicted poverty incidence, is a third indicator that
should be taken into consideration through the process of constructing the
formula; the predicted poverty incidence should be around the actual poverty
incidence. A common approach to evaluate the under-coverage and leakage
rates of a PMTF is a two-by-two table. Consider an actual case where there are
100 households and a poverty line that implies that 20 of these are classified
as poor. And consider a PMTF that predicted 30 households are poor. Of these,
15 are actual poor and 15 are non-poor. Both the 15 poor households and the
65 non-poor households are treated as successful targeting. The 5 predicted
non- poor households are treated as “errors of exclusion, while the 15 actual
non-poor households are seen as “errors of inclusion. While trying to reduce
errors of inclusion, it will also raise errors of exclusion. Similarly, raising the
poverty line in order to reduce under-coverage will also tend to increase
leakage.

105 | I S I W S C 2 0 1 9