Comments on Hallerod (1994a)

Some comments on

Hallerod, B. (1994b) A New Approach to the Direct Consensual Measurement of Poverty.

Because these focus on areas of disagreement they should be seen in context . I like the authors overall approach.

The arbitrariness of the contents of the list: This list should be treated as a menu. The main question then is whether it offers enough variety to enable all possible views of what basic necessities are. Has there been any analysis of how much the size and specific contents of the menu matters ? This could be done by progressively withdrawing 1,2,3... items, during the analysis and seeing what difference it makes to the final size and distribution of poverty .
The use of the label "consensus" is arbitrary in the sense that it is not strictly true. I agree. Very few items are seen as necessities by 100% of those surveyed. I would argue that versions of this approach might be better re-labelled as a democratic approach to the measurement of poverty, for reasons outlined below.
3. "There are no theoretical reasons to put the level for defining necessities at 50 per cent rather than 30 or 70 per cent, or at any other level". I disagree. A "common sense" view of poverty would be one that most people agreed to. 50% or more would equate with common sense. In most voting systems within voluntary organisations, parliamentary elections and meetings of shareholders, 50%+1 of the vote or more is seen as approval. There are good political reasons for using a 50% cut-off limit for defining necessities. The use of this definition would enhance the public acceptance of the results of such surveys. It is also consistent with the intent built into the survey method, of valuing people's views, rather than relying on attempts at expert assessments of basic necessities. Here we are valuing their views about how views should be valued !
A democratic approach has already partly built into the design of the PDI. The weighting system assumes each person's preferences have equal value. There is no sociological or anthropological justification for such a valuation. It is a statement of aspiration rather than reality. It is normative, not descriptive of day to day reality.
"Difficulties arise , however, when, an individual's preferences diverge from the aggregated preferences revealed by public opinion" This is not so much a problem of the measurement approach, but more one of the idea of poverty the tool is trying to measure. People can be seen as poor and feel poor, or be seen as okay but feel poor, or be seen as poor and feel okay, or be seen as okay and feel okay. The bias in M&L's approach is to count only those who are seen as poor and feel poor. This is a tough minded approach to poverty, perhaps too tough. It discounts the fact that people seen as poor by other's may have discounted their expectations of their life as a result of being poor. It also ignores the fact that those people, who may not see themselves as poor, are still likely to be treated as poor by others who see them as poor. We need to be able to differentiate poverty as defined socially and individually. What difference in % terms would it make to your Swedish figures if poverty was recalculated by the absence of publicly defined necessities, not adjusted by people's own preferences for their own life ?
The importance of homogeneity of people's opinions about what were necessities. In M&L report this seemed to be a political sentiment, rather than a theoretical statement, in that it strengthened the apparent legitimacy of the poverty definition used. There was no denial that for many items there was also little agreement.
Adjusting weightings according to the views of appropriate reference groups seems to me deeply problematic, and arbitrary. In extreme, there will always be a reference group where even the most obscure items will be seen as necessities. Demographically defined reference groups are not less arbitrary in their selection than political or economic ones. Analysing differences between groups, after a poverty figure has been calculated by a common weighting scheme seems to be another matter, and entirely worthwhile.

The use of the "don't have and don't want / don't have and don't want" question (part of the definition of the seen to be poor and feel poor) automatically means that very local criteria of poverty are taken into account in the measurement process. Presumably these would be influenced by various reference groups relevant to that persons life.
The definition of the poverty line as the absence of three or more necessities: I thought that this was originally based on an apparent discontinuity in income levels between those in this group and those lacking fewer necessities. However, this rationale assumes a discontinuity in income levels should be found between the poor and the rest of society. There is no reason to assume so. I think the focus on a limited number of items was necessitated because M&L did not use a weighting system for aggregating the absence of various necessities. Without it, their poverty numbers would have been unrealistically high when compared to common perceptions.
I don't think the choice needs to be between the MNI and the PDI, as you suggest on p31. My preference would be to keep the basic weighting system used by the PDI (but not adjusted by preferences of respondents' demographic group) but to include the 50% or more of respondents agreeing an item is a necessity as the definition of a necessity.

Since January 24th 1998 then there have been Visit counter visits to this page.