This page compares two simple methods of measuring poverty:
The comparison was prompted by an email to the MandE NEWS email list by Atta Ullah in May 2008 asking for a comparison.
The similarities between the two methods:
- Both survey instruments are one page long. Both are simple to use in the field.
- Both collect information about observable aspects of people’s lives
- Both are inexpensive to implement.
- Total scores for respondents are easy to calculate, with both methods.
- Both survey instruments produce an aggregate score for households, which is expressed in percentage terms
- Both surveys can be related to national level survey data
The differences between the two methods
Transparency: (as in ease of understanding)
The BNS is based on specific definition of poverty, which emphasises the importance of a democratic definition of poverty (an item is a basic necessity if 50% of more or respondents say it is), and a rights view of poverty (see next paragraph). In contrast, most national surveys, which the SPS would be based on, depend on expert definitions of poverty which are then legitimatised by governments though their use in official surveys.
The definition of poverty in the BNS is easy to explain to ordinary people: “Poverty is the lack of basic necessities. Basic necessities are those things that everyone should be able to have and no one should have to go without“. The definition of poverty in the SPS is derived from the definition in the relevant national survey which was used to derive weights (“points”) for each of the items in the Scorecard. In some cases this national definition may be simple, like households spending less than $1 a day. But in many cases national survey definitions can be quite complex and hard to explain to non-specialists.
The process of identifying weights for items in the BNS is simple. The weight for each item = the % of people who said in the survey that they think the item is a basic necessity (as defined above). The weight for items in the SPS are derived from a statistical analysis (logit regression) of the association between the presence and absence of each item in the score card with the poor / not poor status of the national survey respondents. Items that are highly associated are given higher weights (“points”), and vice versa.
The BNS poverty score for a household tells you how poor they are. It is a “depth of poverty” measure. People with a 90% score lack many basic necessities. People with a 10% score lack very few basic necessities. The SPS score tells you the probability that a household is poor. A 90% score means it is very likely the household poor, whereas a 10% score tells you they are very unlikely to be poor. The SPS score relates to, but is not the same as, a “headcount” measure of poverty. The SPS measure does seem to be a more difficult concept to communicate to non- specialists than the BNS measure.
The same difference in explainability applies to the the collective measures of poverty, applied to groups of households. With the BNS, the poverty status of a group of households could be described by their average BNS score. This would describe the average level of deprivation of households. In the case of the SPS “the share of clients who are poor is the average of their individual poverty likelihoods”. This requires some understanding of probability.
The kinds of items included on the BNS menu are only limited by the need for items to be observable, and for the whole menu not to be unduly long (e.g. longer than one page). The kinds of items in the SPS are limited to those items already included in the national poverty survey they relate to. This is because the national survey is the source of weightings for those items. As with the BNS, the SPS is also limited to observable items and a not-excessive number of those.
The BNS one page menu contains many more items than the SPS one page instrument (approx 25 versus 10). This is because the BNS asks yes/no questions, rather than multiple choice questions about a more limited set of items. Combined with the ability to include locally relevant items on the menu, this seems likely to make the BNS a more sensitive instrument, able to pick up smaller changes in poverty status. But this proposition needs to be tested.
The BNS is based on a specific definition of poverty, which is rights oriented and democratically defined. Basic necessities are those items that everyone should be able to have and no one should have to go without. Items are defined as basic necessities if 50% or more of respondents say they are basic necessities. The SPS is based on expert definitions of poverty as embodied in the national survey used as the source of item weights. Those expert definitions are legitimated by their use in official surveys. The legitimacy of the BNS is based on the values embodied in the method (rights and democratic process).