A preference-based theory unifying monetary and non-monetary poverty measurement

A Modification to the Alkire-Foster Method When Data Are Cardinal

“…When k = 2, which is the in-between case, the numbers of poor people identified by the two methods are relatively close (H M M = 0.200 vs. 0.225 = H AF ). Overall, as the poverty cutoff grows from the union case to the intersection case, H M M exhibits smaller dispersion than H AF , which fits the prediction we discussed in Section 3.1 and exhibits a more stable behavior, showing tendency of reducing misidentification errors according to Decerf (2023). The other measures including M 0 , M 1 and M 2 show the same pattern.…”

“…However, that does not contradict the argument that the identification result is vital for the performance of poverty alleviation programs. Decerf (2023) described two kinds of 'misidentification errors': if a person is poor but not identified as poor, it is an 'exclusion error'; if a person is not poor but identified as poor, it is an 'inclusion error. '…”

“…Naturally, we assume the upper reference is higher than or at least equal to the dimensional cutoff, i.e., ζj ≥ zj .19Decerf (2023) argued that if the preference-based definition of the poor is introduced, then the ultimate goal of choosing appropriate identification criteria seems to be the minimization of misidentification errors or the closest approximation by the identification contour to the real indifference curve.20 More discussion about unidimensional and multidimensional approaches can be seen inAlkire and Foster (2011b).…”

“…The adding and trimming effects could be regarded as remedies against the exclusion and inclusion errors, respectively.24 Such adding and trimming effects behavior is beneficial according toDecerf (2023) because the union criterion tends to cause inclusion error and the intersection criterion tends to cause exclusion error.…”