Beyond Wealth Ranking: The Democratic Definition and Measurement of Poverty

A Briefing Note prepared by Rick Davies (CDS Swansea for the ODI Workshop "Indicators of Poverty: Operational Significance", to be held on Wednesday, 8 October 1997 in London.

TWO IMMEDIATE PROBLEMS

TWO WIDER PROBLEMS

BASIC NECESSITIES SURVEYS (BNS): OUTLINE OF A METHOD

This method of poverty assessment is based on an approach used in Britain in the early and late 1980's by Mack and Lansley (1985; Frayman, 1991), but with an important adaptation: the approach to weighting and how a poverty line is defined.

Poverty is defined as "the lack of basic necessities". Unlike many approaches to poverty assessment there is no a priori definition of what are basic necessities. They are defined during the survey process itself. (See below.)

The starting point is a list of items (e.g. TV, bicycle, radio) and events (e.g. one days holiday per week, all school age children attending school) that people may or may not believe are basic necessities. It is important the list includes items almost everyone would agree with, and others where people may be in much more disagreement. The list should be seen as a menu of possibilities.

This initial list can be generated by wealth or well-being ranking exercises, or focus group discussions.

A sample survey is then carried out using a survey instrument which has three sets of questions

1. Name of respondent and details of household members [basic identifiers]

2. "Which of the items on this list do you think are basic necessities which everyone should have and which no one should have to go without ? [List of items mentioned above, not exceeding 25 in total. These can be read out one at a time, or put on cards and given to the respondent for sorting]

3. "Which of these items does your household have now ?" [Same list of items mentioned above, not exceeding 25 in total. These can be read out one at a time, or put on cards and given to the respondent for sorting]

Results for each household sampled can be recorded in the form of a table, given below.

Respondent Details
Name Age Sex
Household size, etc

Date
Interviewer

Items

Are necessities ?
(0 = No, 1 = Yes)

Household has now ?
(0 = No, 1 = Yes)

1.
2.
3…

ANALYSIS

  1. For each item, identify the percentage of all respondents that believe that item is a basic necessity. We can call that percentage figure the weighting for that item.
  2. Basic necessities will be deemed to be all those items where 50% or more of the respondents believe that the item is one which everyone should have and which no one should have to go without.

    These procedures move us towards a democratic definition of poverty

  3. Poverty Scores for individual households can be calculated as follows:

    For each item now deemed to be a basic necessity:

    Item 1 weighting x "possession score" (0 or 1) =

    Item 2 weighting x "possession score" (0 or 1) =

    Item 3 weighting x "possession score" (0 or 1) =

    Item 4 weighting x "possession score" (0 or 1) =

    etc.................................................................=

    Total..............................................................=

  4. Calculate the maximum possible score for the whole set (assume a person has all of the items).
  5. Divide the respondents total score by the maximum possible score

INTERPRETATION

A person possessing none of the items that are deemed by the respondents to be basic necessities will have a score of 0.00 By the survey respondents' standards this person will be living in extreme poverty

A person having all of the items deemed to be basic necessities will have a score of 1.00

A graph can be produced to show the distribution of the poverty scores for all the sampled households. This will show:

ADDITIONAL FEATURES

1. Standards of what are believed to be basic necessities are likely to change over time if a society is undergoing some form of economic development or cultural change. When repeat BNSurveys are carried out within the same community it is possible to take such changes into account without losing continuity and comparability of poverty data over time Additional items now thought to be more important can be added to the list of items presented to respondents. Scores can be calculated for each person on the basis of the new extended list and then, after excluding those new items, according to the old set. This is similar to how consumer price indexes are updated. The same approach can be taken to the deletion of items that are now seen as redundant because 99.99% of all households now have those items (e.g. inside toilets in the UK).

2. Differences can exist between our view of our needs and other people's views of our needs. We may feel poor but others may not think of us as poor, or others may think of us as poor but we may not feel ourselves to be poor. Our awareness of our standard of living may be mitigated or accentuated by what we know others think of our situation.

By asking the respondents two additional questions it is possible for the BNS to further differentiate the nature of poverty in a community, taking into account differences between how individuals see their situation, and how the surrounding society sees it. We can ask (using a card sorting exercise):

  1. Which of these items do you feel are a necessity for your household ?
  2. Please rank these items (identified as a necessity), according to which is most-to-least necessary, from your point of view.

We can then create an Poverty Score (Self-perceived), as follows.

  1. Firstly, create weighting values for each item. Imagine that there are 8 items, ranked 1 to 8 in importance, and a ranking of 1 is most seen as most important (necessary). To create a weighting for an item given a rank of 1, reverse the ranking value (to 8). For an item given a ranking of 7 the weighting would be 2. For an item given a ranking of 4 the weighting would be 5. To simplify this task list the rank orders possible in one column, then in the next column write out this list in reverse order.
  2. Then, for each item now deemed to be a basic necessity:

    Item 1 weighting x "possession score" (0 or 1) =

    Item 2 weighting x "possession score" (0 or 1) =

    Item 3 weighting x "possession score" (0 or 1) =

    Item 4 weighting x "possession score" (0 or 1) =

    etc...............................................................… =

    Total............................................................ =

  3. Calculate the maximum possible score for the whole set (assume the respondent had all the items).
  4. Divide the respondents (actual) total score by the maximum possible score.

A person who is poor in their own eyes will have a score close to 0.00. A person who is not poor in their own eyes will have a score close to or equal to 1.00.

A (scatter-plot) graph can then be created plotting each individual surveyed according to their Individual Poverty Scores defined by others (IPS-O) and by self (IPS-S). The four quadrants of this graph can summarised in tabular form as follows:

PovertyScore (Self)

(>0.50) "Poor Rich" "Poor Poor "
(<0.50) "Rich Rich" "Rich Poor"
(<0.50) (>0.50)
Poverty Score (Others)

A "tough" approach to the measurement of poverty in a community would only count the incidence of households that were "poor poor " in the above table. This approach would ignore or choose to deny the fact that reduced expectations about standards of living were a part of the experience of poverty.

SOME SUGGESTIONS BASED ON EXPERIENCE TO DATE

1. Carry out a pre-test of the method using a small number of households.

2. Where possible, try to relate the results with other sources of data on relative wealth and poverty. The result should not be grossly contradictory.

3. When working in second languages take great care in translating the questions. Try to ensure that the notion of a right, or entitlement,or legitimate expectation is communicated. Once translated by one person then have that translation back-translated into your language to see what has happened to the meaning of the question.

4. Make sure the list of items contains a diversity of items, not just those most would agree are basic necessities. The pre-test will show the gaps, where further items are needed. It is important to construct a "menu" of items that provides an adequate range of choice for everyone.

5. Make sure the list of items does not only include physical things, but also events (which may cost money). This will enable the more social dimensions of poverty to be included. It is also possible to include events that relate to use of public services or resources, not just those available on the market.

6. Take care that the list does not include items that are vaguely defined. This will cause variations in response simply due to differences in interpretation.

Please Note: If anyone is interested to test/use this method in other locations in Africa or Asia, please inform Rick Davies, at rick@shimbir.demon.co.uk or phone/fax 44 1223 841367.

References:

Cox, A (1997) Letter to participants in an ODI Workshop on "Indicators of Poverty: Operational Significance", to be held on Wednesday, 8 October 1997 in London..

Frayman, H (1991) Breadline Britain 1990s. Booklet by London Weekend Television.

Gordon, D., Pantazis, C (1997) Breadline Britain 1990s. Ashgate Publishers Ltd. UK.

Hallerod, B. (1994) A New Approach to the Direct Consensual Measurement of Poverty. Social Policy Research Centre Discussion Paper No. 50. University of New South Wales. Here are some comments on the contents of this paper which I have sent to the author.

Hallerod, B. (1994) Poverty in Sweden: A New Approach to the Direct Measurement of Consensual Poverty. UMEA Studies in Sociology No. 106. Umea University. Umea.

Mack, J., and Lansley, S. (1985) Poor Britain Allen and Unwin. London.

This paper was last edited on October 11, 1997

An addiitional bibliography has since been provided by David Gordon at Bristol

Mack, J and Lansley, S (1985), Poor Britain. George Allen and Unwin Ltd, London.

Callan, T, Nolan, B and Whelan, CT (1993), Resources, Deprivation and the Measurement of Poverty, Journal of Social Policy, 22(2), pp.141-172.

Hallervd, B (1995), The Truly Poor: Indirect and Direct Measurement of Consensual Poverty in Sweden, Journal of European Social Policy, 5(2) pp.111-29.

Muffels, R and Vreins M (1991), The Elaboration of a Deprivation Scale and the Definition of a Subjective Deprivation Poverty Line, Paper presented at the Annual Meeting of the European Society for Population Economic, 6-8 June, Pisa.

Muffels, R, Berghman J and Dirven, H (1992), A Multi-Method Approach to Monitor the Evolution of Poverty, Journal of European Social Policy, 2(3), pp.193-213.

Kangas, O and Ritakallio, V (1995), Different Methods -Different Results? Approaches to
Multidimensional Poverty, Paper presented at the ISARC 19 Conference, Pavi, Italy.

Mayer, SE and Jencks, C (1988), Poverty and the Distribution of Material Hardship, The Journal of Human Resources, XXIV.1, pp.88-113.

Nolan, B and Whelan, CT (1996), Resources, Deprivation and Poverty. Oxford, Clarendon Press.

Van den Bosch, K (1998) Perceptions of minimum standards of living in Belgium: Is their a consensus? In Andre_, H. J. (Ed) Empirical Poverty Research in a Comparative Perspective. Ashgate, Aldershot, pp135-164,.

Veit-Wilson, J (1987), Consensual Approaches to Poverty Lines and Social Security.  Journal of Social Policy, 16(2), pp.183-211.

Late note: ActionAid Vietnam have now published a report on the results of their use of this method in Vietnam. For copies of this report, contact William Smith, ActionAid Vietnam by email at Wiliams@aaviet.netnam.vn]


Return to Rick Davies - Social Development Consultant or MandE NEWS, or email Rick Davies at rick@shimbir.demon.co.uk


Mande NEWS: Beyond Wealth Ranking moved to this new address on January 19th 1998. Since then there have beenVisit counter visits to this page.