OVERCOMING THE OBSTACLES OF IDENTIFYING
THE POOREST FAMILIES
Overcoming the Obstacles of Identifying the Poorest
Families: Using Participatory Wealth Ranking (PWR), The CASHPOR House Index
(CHI), and Other Measurements to Identify and Encourage the Participation of
the Poorest Families, Especially the Women of Those Families
Anton Simanowitz, Institute for Development Studies, Sussex, Uk
Ben Nkuna, The Small Enterprise
Foundation, South Africa
in Abidjan, Côte d’Ivoire, 24-26 June
1999 and updated June 2000 for the Microcredit Summit Campaign regional
meetings in Africa, Asia and Latin America. The paper is intended to
further the Microcredit Summit Campaign’s learning agenda. The opinions
expressed are those of the authors.)
Executive Summary
Introduction
Cashpor Housing Index
Participatory Wealth Ranking
Checklist Poverty Tools
Conclusions
Annex 1
Annex 2
Return home
It is not a question of cost or
sustainability (although this has a major impact on how poverty targeting is
done). Whether a program is exclusively targeted or not, experience has
shown that to reach the poorest we must specifically design a program that
caters to their needs. Poverty targeting can assist this process by
raising awareness of the different needs of different types of clients and
allowing for different products to be effectively targeted.
Many
people argue that it is impossible, or too expensive, to design reliable
poverty-targeting tools. However, there are a number of cost-effective
screening methods in use. This paper describes three approaches to
poverty-targeting that are effective in identifying the very poor, and which
have been operationalized and utilized on a large scale with thousands of
potential clients.
The
CASHPOR House Index (CHI), uses external housing conditions as a proxy for
poverty, and can be very effective in conditions where there is a consistent
relationship between poverty and housing conditions. Participatory Wealth
Ranking (PWR), uses a community’s own definitions and perceptions of poverty,
and employs rigorous cross-checking methods to ensure consistency and accuracy
of results. Both methods aim to build on existing information, collect
the minimum data necessary for reliable targeting, and follow-up targeting with
a motivation process to encourage the poorest to join the program.
Both
methods are context specific. PWR relies on detailed knowledge of a
community of itself, and is unlikely to work in contexts where the community is
weak, or where there are high levels of conflict or mistrust. Similarly,
the CHI relies on there being a strong correlation between housing conditions
and poverty. This is not a universal relationship and is very much defined
by the context. Where the CHI is adapted to local conditions, perhaps even
including other externally visible, non-housing indicators, there is a greater
chance of the Index being applicable to a wider range of contexts.
A
third tool, we term a “check-list” approach, builds up a list of poverty
proxies or indicators, based on a local understanding of poverty. Scores
are then assigned to each indicator, or a poverty-line level
determined. The poverty level of a household can then be calculated from
their total score, or number of qualifying indicators.
These methods must not be applied blindly but adapted to local needs and conditions. A number of choices need to be made which will determine which tool is used.
Return to top
OVERCOMING THE OBSTACLES OF IDENTIFYING
THE POOREST FAMILIES
Anton Simanowitz and Ben Nkuna, The
Small Enterprise Foundation, South Africa
...When we launched our
program our heart was to reach the poor...and help them to get over the poverty
line...we decided that the method we would use is we would offer a very small
loan size because, surely, only the poor would take a small loan size. The
next thing we did is went to one of the poorest areas in South Africa...and we
began to do the loans. But after a few years, we realized that [of] the
people we were serving, the majority did not live below the poverty line...
Now the clients that are
very much poorer, why don’t they come to you in large numbers? And the
reason is, they’re intimidated by the wealthier clients. What we’ve heard
from the literature from all over the world is what we found in our own case,
and through hard experience. The poorer people see who goes to your
program, and they just say, ‘This program is not for us; it is for those better
off people.’ And then very often the wealthier – maybe just the less poor
– intimidate the poor, simply by saying, ‘This meeting is for serious people.Here
we have to be serious about business. Somebody who is only selling a few
vegetables is not serious about business.’ Poor people already have pretty
low self-esteem, but you add a few comments like that, and they leave. So,
the presence of the non-poor unfortunately did scare away the poor. And
that’s why we have to go for an exclusive poverty focus.
John de Wit
Managing Director,
Small Enterprise Foundation
The Don't Do It Approach to Poverty-Targeting
Why We Should Actively Target the Poor
Should We Exclusively Target the Poorest?
Cost-Effective Poverty-Targeting
INTRODUCTION
Microfinance
has proven to be an effective and efficient mechanism in poverty reduction the
world over. The 1997 Microcredit Summit[1] declared
as its goal to reach “100 million of the world’s poorest[1]
families, especially the women of those families, with credit for
self-employment and other financial and business services by 2005.” This
is a bold objective, since reaching the poorest families through microfinance
is still in its infancy, and most microfinance institutions (MFIs) currently
reach the poor, not the poorest.
This
paper is about the first step in this objective: identifying the poorest
clients. It is a step mostly avoided or forgotten in the clamor to open
up programs that can start dispersing loans, and lose no time in reaching
financial self-sufficiency. Our question is how can microfinance benefit
the poorest if we don’t know who the poorest are? How can we say we are
reaching the poorest if we are not measuring this? How can we identify
these families on the ground, and encourage their participation in microfinance
programs? And how can we measure impact if we don’t know where clients
start?
Early
in the Campaign, it became clear that it would be difficult to track progress
towards the Summit’s goal with the current knowledge in the field. Most
microfinance practitioners can report on their numbers of clients, and the
percentage that are female, but are unable to document how many of their
clients were among the “poorest” when they joined the program. Most
practitioners simply do not have a simple, low-cost method for assessing the
poverty level of their clients.
But
poverty-targeting is more than just knowing who we are reaching, ensuring that
we reach who we want to reach, and reporting to our stakeholders on this. Can a
microfinance program be designed which will attract only the poorest? Or
can we succeed in persuading the wealthier people not to join, and attracting a
mixture of the poorest plus the poor? Or will we end up designing a
program that does not attract the poorest and serves only the poor and the
non-poor?
We
argue that unless active poverty-targeting is used then we cannot build
microfinance services for the poorest. Many programs exclusively target
women, in part because dominance of men can discourage women’s
participation. Similarly, experience has shown that if better-off people
are included, this may well discourage the poorest from joining! Hence,
even if our aim is not to exclusively reach the poorest, unless we use
active targeting we may well inadvertently miss the poorest altogether.
It
is not a question of cost or sustainability (although this has a major impact
on how poverty targeting is done). For us, if we want to reach the poorest
through microfinance, we must specifically design a program that caters to
their needs.
Before we go on to discuss how to
target, it is important to address some of the arguments against targeting.
THE “DON’T DO IT” APPROACH TO POVERTY-TARGETING
The most common arguments against active poverty-targeting, are a somewhat fudged mixture between “it costs too much to do” and “it’s not necessary.”
The
most popular argument against targeting revolves around the issue of costs.
Bringing microfinance services to isolated rural villages is costly. The
proponents of this view argue that targeting escalates transaction costs, thus
undermining institutional financial self-sufficiency (IFS). A focus on the
poorest also means a focus on the least profitable clients, who (initially)
take the smallest loans.
“At a given point in time [MFIs] can either go for growth and put their resources into underpinning the success of established and rapidly growing institutions, or go for poverty impact...and put their resources into poverty-focused operations with a higher risk of failure and a lower expected return” (Hulme & Mosley, V1, p.206).
Active
targeting is also viewed by some practitioners and academics as
unnecessary. This group contends that the objective of reaching the poorest can be
achieved through the design of credit methodology. For example, a
program might offer small
loans with high transaction costs, in terms of time spent to enter the program
and during meetings, in order to deter all but the very poorest from
joining. These design features serve to make the product
unattractive to better-off people.
Another
design choice intended to focus a program on serving the poor and poorest is to
work in areas where most of the poor live. Most countries have crude (and
some more sophisticated) demographic data that will give a rough idea of where
there are higher or lower concentrations of poor people.
Where
these approaches are criticized as unreliable, the response is normally that
this type of passive screening is the best compromise, given the high costs of
more pro-active targeting methods.
It
is our assertion that this is not a compromise, but a fundamental flaw in
providing financial services for the poorest. Experience in the
poverty-focused MFIs demonstrates that active poverty-targeting is a crucial
first step in building an effective and efficient institution for poverty
reduction.
WHY WE SHOULD ACTIVELY TARGET THE POOR
An
institution that is working with the poorest and poor must include activities
that are specifically designed for these target groups – targeting is just one
of these. While this undoubtedly increases transaction costs, there is increasing
evidence that client growth and other benefits from targeting lead to a
progressive movement towards IFS. Innovation and creativity can strengthen
this. For example, targeting can assist in raising awareness and in
motivating people to join the program (this is discussed later in the House
Index and Participatory Wealth Ranking descriptions). Furthermore, recent
innovations in targeting have made it more cost-effective.
Targeting is viewed by many institutions as an essential prerequisite for creating and maintaining good credit discipline, which is crucial for the program to be sustainable. Thus poverty-targeting is not just an issue of reaching the right people, but is a question of creating a functioning structure that will develop into a sustainable institution. Failure to target the poorest and the poor can lead to a lack of focus and a mismatching of services to client needs. This may lead to poor credit discipline, high drop-out rates, and generally create a problematic environment which will eventually undermine sustainability.
There are two clear benefits of
having an understanding of who we are reaching with our financial
services. The first concerns the MFI and its ability to target its
services to desired target clients. Without knowledge of the client
poverty-profile it is impossible to know if we are reaching whom we want to
reach, and to know whether our products and services meet the target clients’
needs.
The second benefit concerns the need
for transparency and accountability in microfinance. Where a poverty
objective is set, it is not good enough to just assume that it has been
reached. MFIs, their funders, and other stakeholders need to have basic
information to describe the poverty level of the clients being reached.
In
this paper we argue that using a poverty-targeting tool is about more than reporting
on figures of who is being reached. It is also a fundamental requirement
of designing an institution that is focused on the needs of the target group,
rather than being pushed by the dominant voice of non-target beneficiaries.
Passive targeting through product
design will always be a doubtful method in reaching the poor, and will
certainly not separate the poorest from the poor. There are
numerous examples of how such loan products have benefited better-off clients
who join with the expectation of larger future loans. The opening quotation in this paper describes
the experience of the Small Enterprise Foundation (SEF) in South Africa, where
it was found that lack of targeting had pushed the program towards serving
better-off clients. Similar experiences have been reported from Asian
integrated rural development programs (Kasim, personal experience).
Poverty targeting is the first step in creating a program that is designed to meet the needs of the poorest. The marginalization and social exclusion faced by the poor mean that a “poverty culture” must be created. By choosing the characteristics of the people we wish to target we are also targeting the program’s benefits.
A
focus on the poor usually goes hand-in-hand with a focus on women. There
are strong arguments that unless programs are directed at women, the poorest
women can easily be bypassed. Women are often targeted because it is
believed that this leads to greater impact in terms of poverty-alleviation, as
women are more likely to spend surplus income on the needs of their
families. In her study of Grameen Bank replications, for example, Helen
Todd described significantly greater impact on women and their families, where
loans are taken by poor women, rather than by other people.
Women
may also be targeted because of their relative marginalization and higher
incidence of poverty. A number of studies have examined the role
microfinance plays in empowering women, but this is subject to much debate by
people who argue that microfinance may contribute to, but does not
automatically lead to, women’s empowerment (see Johnston; Johnston and Rogaly;
Mayoux).
A
third common reason for the targeting of women relates more to the operational
benefits than to poverty alleviation. Experience has shown that very poor
women perform much better than men in terms of loan utilization and credit
discipline. Rutherford (quoted in Johnston and Rogaly, p.14) notes that in
Bangladesh women are targeted because they “are seen as more accessible (being
at home during working hours); more likely to repay on time; more pliant and
patient than men; and cheaper to service.”
SHOULD WE EXCLUSIVELY TARGET THE POOREST?
The Summit’s goal is to reach 100
million of the world’s “poorest” families. The reality of
microfinance today is that there are few organizations that really concentrate on reaching
the poorest.
The largest factor driving this is
financial self-sufficiency. In the discussion above we argue that creating
an institution for the poorest does impose additional costs on an MFI, but it
also creates benefits that can lead to IFS. We also argue that it is only
in squarely facing up to the challenges of building financially sustainable
MFI’s for the poorest that we will overcome the cost challenges.
Thus the exclusive targeting debate
has at its core the ability of MFIs to reach and serve the poorest. This
is a critical question that must be central to the Summit’s goal of reaching
the poorest.
Pragmatism would tend to favor an
approach that accepts that most MFIs will reach a mixture of clients, including
some of the poorest and the poor, but perhaps also the non-poor. In this
case, poverty-targeting would be mostly a tool for understanding and reporting
on who is being reached. Provided that the poorest can be effectively
reached in a “mixed” program, increasing scale is likely to lead to significant
outreach to the poorest. Programs serving several strata of clients, not
just the poor and the poorest, may be able to expand faster and to reach larger
numbers than programs exclusively targeted to one stratum. If they do, large
numbers of the poor and the poorest are likely to benefit. Moreover, such
programs have the possibility of cross-subsidizing their less profitable
lending to the poorest from their more profitable lending to the non-poor, and
thus potentially achieving IFS more rapidly.
There is no intrinsic problem with a
program targeting both the poorest and the poor, and even the
non-poor. However, where the needs of the different target groups are very
different, it becomes more difficult to meet the needs of the poorest.
Organizations, such as SEF, which have
tried to work exclusively with the poorest, have found that, in their context,
the poorest cannot be effectively reached in a mixed program. To meet the needs
of the poorest, a culture of poverty-focus needs to be created. If there
is a dominance of non-poor, or even poor over the poorest, then the stronger,
more confident, more vocal people will make their voices heard. An
innovative MFI striving for IFS or a loan officer responding to financial
incentives for a good loan portfolio will listen to these voices. This
will lead to a tendency to develop loan products more suited to the better-off
clients. Poorer clients become increasingly marginalized both by the MFI,
and in the case of group-based lending, by their fellow group and center members. The
very poor, who take small loans, experience problems, are vulnerable and need
much support, and are not popular with other clients, loan officers, or
branches striving for profitability. At such a point, if the program is
still attracting the very poor, it may well lead to negative impact for these
clients, and in fact contribute to greater poverty.
Whether a program is exclusively
targeted or not, experience has shown that to reach the poorest we must design
services to meet their needs. A program can reach different market groups,
but must understand their differing needs and treat each group
accordingly. Poverty-targeting can assist this process by raising
awareness of the needs of different types of clients and allowing for products
to be effectively targeted.
We therefore need to target the
people who make sense in terms of program design both from the perspective of
providing a service that meets the needs of the poorest, and that allows for
the development of an MFI that can achieve IFS. The exact composition of the
target group is therefore defined by a combination of recognizable groupings of
people at different poverty levels, the extent of the differences in services
needed for these different groups, and the practical demands of establishing a
viable MFI which can achieve IFS.
COST-EFFECTIVE POVERTY-TARGETING
If we accept that we need to actively target the poorest in order to ensure that they
· join the program,
· stay in the program and are not pushed out by others, and
· create an efficient, effective, and cost-effective institution that is shaped to the needs of the poorest
then
we can start to look at how this can be achieved in the most cost-effective
manner.
This
paper describes three approaches to poverty-targeting that are effective in
identifying the very poor, and which have been operationalized and utilized on
a large scale with thousands of potential clients. Targeting tools develop
simple methods to measure “poverty”, which is complex, subjective and very difficult
to measure accurately. Compromises have to be made.
Three key questions face MFIs that want to develop poverty-targeting tools:
1) How do we simply measure a complex issue such as poverty?
2) How can we be confident about the results that we obtain?
3) How
can we make sure that the targeting tool is cost effective?
1)Creating
a simple tool to measure complex poverty
1) Using
community definitions and self-ranking: Tools based on this approach
acknowledge the complexity and subjectivity of poverty, and assert that
“insiders” are the most knowledgeable, and use communities own definitions of
poverty. The design challenge is to find a way of obtaining consistent and
honest information from communities. Participatory Wealth Ranking (PWR)
is the most commonly used example. It uses a community’s own
definitions and perceptions of poverty, and employs rigorous cross-checking
methods to ensure consistency and accuracy of results.
2) Visual
indicators of poverty: By understanding local conditions and
characteristics of poverty it is possible to select one or more indicators or
proxies for poverty that are visible during a short visit to a person’s
house. Where a visual indicator has a strong relationship to poverty
level, this can be an effective an low cost approach. The Cashpor House
Index (CHI) uses external housing conditions as a proxy for poverty, and
can be very effective in conditions where there is a consistent relationship
between poverty and housing conditions.
3) Check-list
of indicators: A number of organizations have developed a “check-list”
of poverty indicators or proxies (ranging from income/expenditure, land or
assets, to health and education or access to water). These tools are based
on an understanding of the local poverty context, and may or may not include
community definitions of poverty, or visual indicators. They develop a
list of indicators that are scored (and sometimes weighted) and thus
triangulate each other. The check-list approach may be a powerful
targeting tool where sufficient time and resources are put into developing the
right mix of indicators. They are generally more time consuming and
expensive to implement than community ranking or visual indicators.
2)
Creating a rigorous and reliable tool
Once
a means has been devised to measure poverty through one of the simple
approaches above, the next consideration is the development of procedures to
ensure that the tool produces consistently accurate results, that are not open
to distortion by staff or clients, or by frequent “exceptional” cases.
A
key innovation in most targeting tools is to build in triangulation. Information
is collected from a number of different perspectives. As each source of
information cross-checks that from others, the confidence in its reliability
increases.
In
community self-ranking methods it is important to ensure that the community
representatives are consistent in their application of the criteria they use,
and that there is no favoritism or attempts to manipulate the results.
Where
visual indicators are used, it is important that a very strong relationship
between the visual indicator and poverty is established, and that there are not
frequent “exceptional cases”. Good systems for appeal and for monitoring
the effectiveness of the system are important, and strong base-line work must
test the reliability of the proxy prior to its use.
Similarly,
in check-list systems there must be good base-line work and a period of
refinement to check which indicators are most sensitive, develop appropriate
weighting and effective triangulation.
3)Ensuring
cost-effectivness
Finally,
for an assessment tool to be useful it must be inexpensive or cost-effective to
use. There are some general principals and approaches that are common to
CHI, PWR and other poverty assessment tools used, which help to transform the
tools from academic exercises to practical methods that can be used on a
day-to-day basis by MFIs.
MFIs
do not need to be exact in their poverty assessment. They need to know
with reasonable confidence the approximate and relative poverty of their
clients compared to the rest of the community. No poverty-targeting tool
can be 100% effective, and the cost of obtaining the last few percentage points
usually outweighs the benefits. In PWR, for example, those last few
percent can easily double the costs of the whole exercise! Thus, a tool
should be based on the minimum amount of information needed to achieve the
minimum level of accuracy desired: “optimal ignorance”.
Starting with Existing Information
The first step
for poverty targeting is to look at a wider level. A province, district,
division, county, village, or other administrative division can be selected
according to available data on poverty levels.
Information
may be available through national or local government departments, dealing with
poverty-related issues such as health, social welfare, education, or
agriculture. Non-government organizations working on related issues can be
solicited for their knowledge of the location of major concentrations of poor
households in a district.
Demographic
data, particularly related to income and expenditure, is notoriously sparse,
and where available, unreliable, especially in developing countries. This
type of data needs to be used very cautiously and only down to the level where
it is reliable. Thus, in some countries this may help to identify the
province in which to base a program. In other countries it may be possible
to target down to the village level using existing data sources.
In
CASHPOR’s Malaysian experience, national statistics are heavily relied
on. Full advantage is
taken of existing statistics on the geographical distribution of poor and
poorest households. Often these are available from recent censuses or sample
surveys and can give the total number of poor and poorest households, or an estimate
of them, down to the administrative district/county level. This allows CASHPOR
to start working in the district/county with the largest number of poor and
poorest households. The next step is to meet well-informed district/county
officials to ask in which villages are the largest concentrations of poor and
poorest households located.
Once these villages have been identified, targeting on the ground can commence. The task begins with visits to the village with the highest number of poor households first, and then to the village with the second highest concentration down, and so on.
Motivation
of eligible people to join the program is an essential part of both the CASHPOR
and SEF methodologies. Although not explicitly part of the
poverty-targeting tool, it is an integral part of their targeting
methodologies.
Many
very poor, especially women, will be too afraid at first to come forward to
join, as they will not know
nor believe that the services are actually for them. Even when informed, many
are likely to feel that it would be too risky for them to borrow. Only
patient and persistent motivation work among them, and the convincing
demonstration effectfrom neighboring poor and poorest households that do
participate and benefit will encourage them to take advantage of the
opportunity.
The House Index was
developed by CASHPOR, a network of Grameen Bank replications in the
Asia-Pacific region, for their members to achieve the twin goals of both
increasing and deepening outreach of poor households and achieving IFS.
There are three steps in the methodology:
1. Identifying high-density poverty areas (discussed above).
2. Using the house and sometimes the compound of the household as a crude indicator to eliminate the obvious non-poor households.
3.
Conducting a more detailed household interview, or Net-Worth test, to determine
program eligibility amongst the remaining households.
How Does the House Index Work?
Sometimes
it is as easy as looking at the roof. In rural Asia, a temporary, flimsy
roof (e.g., straw, leaves, plastic sheets, or cardboard) nearly always
indicates a "very poor" household, as distinct from a
"poor" one. Replacing the roof, for example, with used
galvanized sheets in the Philippines, or locally manufactured, second-hand
tiles in South India, does not cost much, but it is still beyond the resources
of the "very poor." Once people get some surplus income (i.e., once
they are less poor), they tend to replace their roof with at least a
semi-permanent one that keeps out the rain and wind and does not have to be
constantly repaired.
For
the convenience of practitioners, the following classification can be a useful
guide. It would, of course, need to be adapted to suit local conditions.
|
CASHPOR House Index |
|
Adaptation to South India |
|
Adaptation to China |
|
|
Size of the house: |
|
Size of the house: |
|
Size of the house: |
|
|
Category |
Point |
Category |
Point |
Category |
Point |
|
Small |
0 |
Small <20 sq. meters |
0 |
Small |
0 |
|
Medium |
2 |
Medium 20-29 sq. meters |
2 |
Medium |
2 |
|
Big |
6 |
Big >29sq. meters |
4 |
Big |
4 |
|
Structural condition: |
|
|
|
Structural condition: |
|
|
Category |
Point |
|
|
Category |
Point |
|
Dilapidate |
0 |
|
|
Dilapidated |
0 |
|
Average |
2 |
|
|
Average |
2 |
|
Good |
6 |
|
|
Good |
6 |
|
Quality of walls: |
|
Height and materials of walls: |
|
Quality of walls: |
|
|
Category |
Point |
Category |
Point |
Category |
Point |
|
Poor |
0 |
<4 feet mud |
0 |
Poor |
0 |
|
Average |
2 |
4 feet mud |
2 |
Average |
2 |
|
Good |
6 |
>5 feet mud |
6 |
Good |
6 |
|
Quality of roof: |
|
Quality of roof: |
|
Quality of roof: |
|
|
Category |
Point |
Category |
Point |
Category |
Point |
|
Thatch/Leaves |
0 |
Thatch/Leaves |
0 |
None/Mud |
0 |
|
Tin/Iron sheets |
2 |
Tin/Iron sheets |
2 |
Partial stone |
2 |
|
Permanent roof |
6 |
Tiles and other good materials |
6 |
Cement/Concrete |
6 |
A
cut-off score of total points will be established to separate the houses of the
poor from those of the non-poor. If there is doubt about what this point
should be, the cut-off can be validated against household incomes in a sample
study. Then, a cut-off score is established within the poor category to
separate the houses of the poorest from those of the poor. Again a sample
study can be carried out to validate the cut-off score against household
incomes. Houses below the cut-off between poor and non-poor scores are
listed and roughly mapped (drawn up in a diagram) as the field staff move
through the village.
The following cut-off points can be a suitable guide in determining eligibility:
i.3
or less: Likely to be Very Poor
ii. 4 to 6: Poor
iii.
Greater than 6: Unlikely To Be Poor
By
limiting the cut-off point to a total score of 3 or less, we are more likely to
be reaching the poorest households. However, for those with a score of 4
or above, an appeal procedure is developed where the households concerned can
go to a more senior staff and bring their case for consideration, freeing the
field staff to go about identifying the poorest households.
All
houses in the whole village with a total score of 3 or less are marked on a
simple map of the village. The number of "eligible looking
houses" is then aggregated and the summary is transferred into a
district-based listing of estimated very poor households. This can be used
to derive a preliminary estimate of the demand of microfinance services. It
will also determine priority in Net-Worth testing. The houses that obviously
are not lived in by poor households (i.e., large houses in good condition, made
from expensive materials and with large compounds perhaps bounded by fine,
sturdy walls) must be ignored.
The
Net-Worth Test
Once
the list and map of eligible-looking houses has been completed for the village,
the field staff will go to the houses on the list to verify the eligibility of
the households through a short, 10 to 15 minute interview that focuses on the
value of their productive assets. These assets include agricultural land
owned and/or operated, farm equipment (including buildings) and machinery,
large farm animals, transport vehicles, and stocks of goods for sale, etc. The
total value of household assets is estimated, and compared with the locally
relevant cut-off that distinguishes non-poor from poor households and the poor
from the poorest. Households falling into the latter two categories are
informed of their eligibility for the financial services and motivated to form
a group with any four other eligible households in the village that they can
trust in matters of money.
Key
Issues in the Use of the House Index[1]
Verification
and quality control
As
part of quality control to ensure minimum leakage of program services to
non-poor clients, and to verify and monitor field staff, revisiting some houses
and checking on the eligibility score and net-worth tests is highly
recommended. Among Grameen Bank replications, the above process is termed
“re-interview,” and is a regular but independent process of random
checking of the quality of output of all field staff, normally undertaken by
the field-level supervisory staff or internal audit team of the head office.
In
new branches re-interview is done in all eligible households. With
increasing numbers of staff, the re-interview is reduced to a random sample of
five per week per loan officer.
Through
this process, field supervisors can be assured that the field staff have not
missed important factors that need to be taken into
consideration. Feedback to the staff and affected households through
re-interviews supports the quality and transparency necessary to ensure public
confidence in the MFI’s ability to reach and benefit the poorest
households. Re-interviews must be carried out if there is any reason to
suspect that better-off households are infiltrating the program in any
particular area. Only when the field supervisors are convinced that a
household is eligible can official confirmation of eligibility be given.
The CHI is a cost-effective and
powerful tool. However, it is only as good as the relationship between
poverty and housing, where it is clear that living in a bad housing
structure indicates poverty, and a better structure indicates (relative)
wealth. Sometimes, therefore, the CHI may not be effective. For example, where the
poor have benefited from public housing programs, as in some Scheduled Caste
villages in southern India, it cannot be used to distinguish between the poor
and the non-poor.
In some contexts the CHI can be
easily adapted to local criteria (see table
above). In other contexts, where a housing index is not appropriate, it
may be possible to develop other visual indicator poverty tests, using local
proxies that show a strong relationship to the poverty level.
However, the CHI and other visual
indicators will always be limited by their need for externally visible
characteristics, lack of contextualization in terms of changing circumstances
over time, and an inability to consider non-physical aspects of
poverty. This is partly dealt with in the CHI by an appeal process and a
Net-Worth test, but these mechanisms add to the cost of the tool and can only
be used for borderline cases.
Comparing CHI to PWR - A Case Study
of SEF, South Africa
The Tšhomišano (pronounced
“chom-is-ano”) program of SEF adopted a modified House Index as its
poverty-targeting tool. Initially an interview similar to the net-worth
test was used, but this was later dropped to save time. Reports of
problems by field staff raised concerns. These concerns mostly centered on
people who had been denied access to the program on the basis of their housing
condition, despite obvious signs of poverty and the confirmation of this by
members of the community. There were also reports of people joining who
the field staff felt were not poor, but who qualified on the basis of the index.
Two studies compared the results of
PWR and CHI, as it was applied at SEF (Simanowitz). These demonstrated the
inaccuracy of a system based on static, externally judged criteria (House
Index), when compared to a local judgement of poverty (PWR). Many
instances were cited of people living in poverty while having reasonable
housing conditions. For example, there were people living in houses
constructed prior to the main income earner dying or deserting the family. In
addition, up to one-third of people classified as among the poorest by the
House Index were among the top half in the PWR. For example, many people
who were living in poor quality housing (while constructing new homes or having
the main home elsewhere) were falsely included in the poorest category under
the House Index.
This highlights the need to ensure
the effectiveness of the proxies used in a visual-indicator system. A
visual system is by definition static, and anomalies do arise. These are
to be expected in a cost-effective system based on the principal of optimal
ignorance. However, it is essential to have an effective appeal system to
deal with the anomalies. The case study also demonstrates the need for the
second stage in the CHI (i.e., the Net-Worth test), to verify results from the
visual assessment. In SEF’s case, this stage was discarded due to the time
involved, but this weakened the targeting process.
PARTICIPATORY WEALTH RANKING (PWR)
In
contrast to the CASHPOR House Index, PWR aims to draw out local knowledge and
criteria on which to judge poverty. The ranking is based on the subjective
views of the people in a community, who generate their own criteria with which
to rank poverty or wealth. Scoring or ranking is facilitated by field workers,
but is performed by members of the community. Visual factors may play a
part in the assessment of poverty, but community members are free to choose
those criteria that are important to them, which usually includes
socio-psychological factors not visible nor easily accessed without a good
understanding of the community.
How
Does PWR Work?
The
ranking takes place in three parts -- mapping, reference groups, and analysis.
Mapping
Step
1. Mapping: A community meeting is set up involving representatives from
all areas of the village in which the program will be established. After
introductions and explanations, a participatory mapping exercise is
facilitated. A village map is drawn either onto the ground with a stick,
or on the floor of a building using chalk. If the village is large (over
100 households), it is divided into recognized sections. A household list is
generated from the map, with participants writing down the commonly used name
for each household (i.e., not necessarily the household head), and these names
are written onto cards. There is one card written for each household in the
area.
Reference
groups
Step
2. Setting up reference groups: Three reference groups are set up for
each section that has been mapped, with three to five members of the community
in each group. If enough people are not present, those present are asked
to invite additional people to the groups.
Step
3. Card Sorting/Wealth Ranking: Each reference group meets separately
and sorts the household cards into piles, according to wealth along a continuum
of high to low. The number of piles generated is determined by the
participants, but at least four piles should be made. During the process of
card sorting much information is gained about the participants’ perceptions of
wealth and poverty, and there is opportunity for discussion.
A
staff member facilitates the process and monitors whether the ranking is done
openly, with good participation of the group.
Analysis
Step
4. Triangulation: The results of the ranking are triangulated by using a
minimum of three reference groups. The numbers given to each household
from each reference group are then totaled. Consistency between the groups
verifies the results. If more than 10% are either gross inconsistencies or
households with missing data, then an additional reference group is required.
Step
5. Scoring: Piles are scored using the formula [100 divided by the
number of piles then multiplied by the pile number], so that the poorest pile
always scores 100. For example, with four piles the poorest pile would
score [100/4 x 4 = 100], the next [100/4 x 3 = 75], the next [100/4 x 2 = 50]
and the wealthiest [100/4 x 4 = 25]. With five piles the scores would be
100, 80, 60, 40, 20.
The
final score of each household is the average of the ranks it was given by the
three reference groups. For example, if it had scored 100, 75, 80 on the
three references groups, its final score would be [(100+75+80)/3] = 85.
Step
6. Selecting the poorest: By this time, field staff have collected
information about the poverty characteristics of the households in each pile
during the reference group sessions. This information is analyzed and
compared with a checklist of poverty characteristics drawn up from the
information given in a number of rankings.[1] The
ranking score that corresponds to these characteristics is taken as the cut-off
point score. The selection of the poor who can be included in the program
is done on the basis of this cut-off score.
Benefits
and Advantages
PWR
is conceptually simple and the results are transparent.Although no attempt is
made to generalize the findings beyond the community in which the ranking is conducted,
comparisons can be made in an area of similar communities where the ranking
criteria tend to be fairly consistent. While the subjectivity of the
results may create “anomalies” when compared with accepted “objective” measures
of poverty, this is balanced by the increase in the community’s acceptance of
programs that work according to their perceived needs, which affirm the
validity of community-defined poverty criteria. The process generates
increased understanding of the livelihoods of members, their perceptions of
poverty, and the consequences of poverty. This is useful for deciding
where the cut-off point should lie, and assists in designing financial products
and in measuring impact.
Effectiveness
of PWR as a Poverty-Targeting Tool
Evidence
as to the effectiveness of the methodology can be demonstrated in four areas.
1.
Triangulation of results: Integral to the PWR methodology is the
cross-checking of data. With triangulation (having each household scored
by a minimum of three reference groups), a high degree of reliability and
validity can be achieved, and poor results are easily detected. Thus,
while the method may be open to manipulation by participants or staff, the data
is highly sensitive to this, and it would be very difficult for this
manipulation not to create high levels of inconsistency-
leading to the data being rejected and the need for the process to be redone.
2.
Feedback from SEF members: SEF has been using the method since
1996. During this period, feedback from communities and field workers has
been almost always positive. There are obviously some cases of people
being left out, but this is probably below the level of 10% inconsistencies
allowed for in the method.
3.
Comparison with the House Index: Strong evidence for the effectiveness of
PWR comes from its comparison with the House Index (see section above on
Comparing CHI to PWR, and Simanowitz, 1999).
4.
Scaling up to large communities: Participatory Wealth Ranking and mapping
methods are generally used in an intensive way by organizations working in
small areas. An ideal-sized community would have less than 100 households,
enabling both the mapping and wealth ranking to be carried out in a short
period. The scaling-up of this method to be used in very large villages (500 -
1000 households), and to be operationalized over a large area was potentially
problematic. SEF’s experience, however, has demonstrated that the methods
can be successfully used in very large villages, provided that the village is
divided into smaller sections, and provided that people are comfortable in
discussing the poverty status of their neighbors, and where there is a
community where people have good knowledge of one another.
Key
Issues in the Use of PWR
Limitations
and challenges
The
practice is more complicated than the idea, so facilitators need to be skilled
and sensitive. If the method is applied without full understanding,
flexibility, and sensitivity on the part of the staff, then poor results are
obtained and resources are wasted. However, poor results are easily
detected due to the rigorous triangulation of information in the method.
Deliberate
distortion of results by participants can make the results unusable, although
this again is easily detected. Experience has shown that the approach of
the facilitator in introducing and facilitating the process is key to gaining
the trust and cooperation of the community. Even in situations of conflict
common in South African villages, deliberate distortion of the results is very
rare, and it is extremely rare for a reference group to have to be discarded.
The
main challenge, therefore, has been the identification of the most sensitive
elements in the method so that the training and assessment of field staff can
be strengthened accordingly.
Problematic contexts
In some situations it may be
difficult to obtain full or open participation from a community, or the work
involved in creating the trust to do this may make the process prohibitively
expensive. For example, problems have been reported when participants are
reluctant to exclude fellow villagers from participation in the
credit program, leading to high numbers of people being ranked as very poor or
poor.
In
another example from Mirzapur, Uttar Pradesh, India, it was found that the
methodology required a lengthy period (days) of gaining “confidence” from the
villagers before they were willing to “open-up” to discuss sensitive issues
such as who are poorer and who are better-off. In addition, to get very poor
women to sit together with others was itself a major task, as the poorer they
are the more secluded they are socially and culturally.
Good
facilitation is the key to overcoming most problems, but there may well be
contexts where PWR cannot work. However, forms of participatory wealth
ranking have been used by development programs throughout the world.[1]
In SEF’s experience it is important to implement PWR through a learning process
to develop the most effective ways to approach communities, and to facilitate
the ranking process. This approach may well lead to solutions to the sort of
problems described above.
Costs
of PWR
Time
and resource costs
In
SEF’s experience the costs of PWR are approximately the same or slightly less
than the screening stages of the House Index (i.e., before the Net-Worth
test). In addition, the process generates a lot of awareness of the
program in the community, and results in far less motivational work being necessary
by the field staff in order to generate clients - this
saves time and resources.
In
a village of 500 households the following would be a typical time allocation
for a PWR team of a supervisor plus three facilitators. The participants
from the community also contribute time.
Mapping:
Three people @ .5 days = 1.5 person-days
Reference
groups: One facilitator can complete three reference groups in one day; i.e.,
each person can complete one section. Five sections would therefore take
five person-days (the team would complete this in two days)
Analysis
by supervisor: 3 hours
Checking
of results by zonal manager: .5 hours
In
addition, stationery (e.g., flip charts, pens, chalk, etc.) is also
required. Refreshments are normally provided during the mapping exercise
as a break between drawing the map and generating the household list.
The
total cost for SEF to rank a typical village in South Africa is therefore
approximately R300 (US$50) plus 7 person-days.
Time
is also required for making arrangements to conduct PWR. This includes
discussion with the community and making arrangements for the
venue. However, this can be done as part of the process of starting a
program in a new village, and need not be budgeted for separately.
Achieving
cost-effectiveness - Optimal ignorance
A
balance needs to be struck between the level of accuracy required in poverty
targeting, and the resources required to achieve this. Thus a key question
in the operationalization of PWR is to find the point of “optimal ignorance,”
thereby reducing the time and resources required to a minimum.
PWR
very rapidly builds up consistent results for the vast majority of
households. The few remaining inconsistencies, however, require that
repeat reference groups be conducted to achieve close to 100%
consistency. Practice has demonstrated that there are rarely more than
about 10% of households that are not consistently ranked by the three reference
groups. Accepting this margin of error - particularly for those
who obviously do not rank amongst the poorest, allows the number of reference
groups to be kept to three (or four in exceptional circumstances). This reduces
the time required for the process, and it is expected that one field worker
would complete the process for at least one section of a village in one day.
Staff
Requirements
Skill
levels
The
skill level required for PWR is higher than for the House Index. However,
a field worker with an average level of intelligence and education can
facilitate PWR; at SEF most field workers have basic levels of higher
education. PWR does require thinking by the field worker, but it is their
approach, and the way in which the tool is used, that is central to the
process.
PWR
supervisors should normally be at the Branch Manger level or higher.
Training
To
implement PWR on a wide scale, standardization through rigorous training,
assessment, and monitoring of facilitators is essential. However, there is
a real danger that standardization will lead to the step-by-step process being
followed blindly, instead of facilitation through an awareness of the progress
of the process. For example, during the card sorting, there is an
introductory discussion dealing with concepts of poverty and wealth, which is
critical to starting the ranking from a common understanding. At SEF, a
form was developed to outline the questions to ask, and which provided space to
record the information given. Immediately the process was interpreted by some
staff as a questionnaire, rather than a facilitated discussion. Yet, if the
steps are not clearly laid out and understood then staff will omit key
elements, and the effectiveness of the process will be undermined.
Staff
training is, therefore, key. This training needs to be experiential and
ongoing. Staff undertaking PWR are supervised for a period and then
formally assessed. Continued spot checks and refresher workshops are
advisable.
CHECK-LIST POVERTY TOOLS
There are a group of tools used by many organizations
that we call “check-list” tools. These are simplified household poverty
surveys, and are able to develop a list of a small number of indicators that
when combined give a reliable assessment of the poverty level of an individual
household. Five tools have been examined for this paper: The Kabalikat
para sa Maunlad na Buhay Inc. Means Test (KMBI, Phillipines); Rhunu UNESCO (Sri
Lanka); Family Development Fund (FDF, Egypt); International Rescue Committee
SEAD Program (IRC, Ivory Coast); and Lift Above Poverty Organization (LAPO,
Nigeria). The second stage of the CHI – the NetWorth test – is also an
example of a check-list tool.
The innovation in this approach
is not the detailed check-list developed by an individual organization, but the
processes of:
1)
developing an accurate and
reliable check-list, which produces results which reflect the local reality,
and does not produce large numbers of anomalies
2)
developing a method of
applying the check-list which obtains good cooperation by potential clients and
their communities
3)
implementation of the method
in a way that is cost-effective and does not require excessive staff time
4)
developing rigorous quality
control to ensure that the results cannot be manipulated by potential clients
and staff.
The indicators selected very much depend on the local
understanding of poverty, and the identification of key aspects of poverty
where there is a simple proxy that can be measured.
These can be divided into four areas:
1.
Income and Expenditure: Many organizations
ask direct questions about actual household income and expenditure, or income
sources, in an attempt to measure economic poverty. The KMBI Means Test
asks for current income of the applicant, the spouse and immediate family
members. Rhunu looks at income bands, whilst the Family Development Fund
defines a poverty-line income level of US$12 per month, and potential clients
are assessed as below or above this line.
Income and expenditure questions are notoriously
unreliable. They are very open to under or over reporting according to
perceived benefits by the potential clients, and are also open to human error
where a client may not know the net profit from a business, or the income from
her husband. Generally a proxy indicator for income will be more reliable than
a direct question, but income bands may be useful, particularly where there is
triangulation from other indicators.
2. Indicators
ofeconomic status: Because of the difficulty in directly measuring income
and expenditure, most organizations use proxy indicators to give an indication
of household income level. These proxies include household assets such as
furniture, television, fridge etc (used by IRC and KMBI); productive
assets such as land or business value/equipment (used by FDF KMBI and LAPO); or
externally visible assets such as housing (usedby Rhunu, IRC, KMBI and
LAPO). A small number of organizations attempt to measure more general
poverty indicators, such as nutritional and health status (used by Rhunu).
Economic status is an important component of poverty, and
where appropriate indicators are developed they can be effective indicators of
poverty.
3. Social indicators:Recognizing
that poverty is much broader than just economic issues, a check-list system
should also include social indicators of poverty. For example, LAPO looks
at marital status, whilst FDF includes the category of female headed households
as positive criterion for identification of the poor – this includes widows,
divorcees, women married to unemployed or disabled men. Level of education also
fits into this category and is used by Rhunu, IRC, and LAPO.
Household poverty surveys
typically include several pages of detailed questions in order to assess
poverty. This is too costly for MFIs, and a small number of indicators are
used instead. However, check-list tools cannot accurately measure poverty
without a demonstrated strong link between the indicators being measured and
the poverty they are measuring. It is therefore important for more
detailed household surveys or qualitative research to look at poverty in the
communities in which the MFI is working/plans to work, and to select key
indicators for poverty. Community involvement in this process, for example
through a wealth ranking exercise, can increase the reliability of information
and lower the costs involved for the MFI.
There is often a tendency to
include too much information, and to “just ask one more question”. But
each indicator included increases both the cost of developing the tool and its
implementation, so the rule of optimal ignorance is vital.
From the examples given above it is clear that although
the five tools bear many similarities, they vary significantly in the specific
indicators they use and the mix between economic, social and wider poverty
factors.
The key is to develop a sound understanding of local
poverty characteristics, and to select indicators that bear a strong and
consistent relationship to poverty level. This is very much something for
each organization to do itself, and the specific indicators cannot be
recommended in this paper. “Best” indicators are those that are directly related
to poverty level, give reliable results, and can be simply and cheaply
measured.
Measurement of indicators
Check list tools use simple and quick ways of obtaining
information. All the tools reviewed include a one-to-one household
interview, but this can be simplified by asking for information that is easily
answered (e.g. household assets rather than actual income), and by asking
information bands rather than exact figures. Visual indicators are also a
simple and cost-effective indicator and can be usefully included.
A method is then designed to
survey potential clients by way of interview and, in some cases, direct
observation, to score them according to a check-list that will rank potential clients
according to a predetermined scale. The check-list will typically be
implemented by a loans officer who will visit the potential client in their
home. She will make observations during this visit, for example of housing
condition and household assets, and this will be included as part of the
poverty assessment. She will also ask a short questionnaire, lasting
perhaps 10 minutes.
Check-list tools use a number of indicators that are
cross-checked or triangulated against each other. The more indicators used, and
the greater the mix of the types of indicators, the more complex is the task of
determining a final poverty score and comparing the scores of different
people. All the tools reviewed, except for FDF use a point system
assigned to each indicator. Rhunu and LAPO weight the points so that each
indicator may have a different number of possible points allocated to
them. The point-based systems then develop a score sheet, and assign
poverty levels to different total point scores.
FDF is simpler in that it uses
qualifying indicators, for example if the family has an income below US$12 per
month they qualify as below the poverty-line. Where a person qualifies in
a certain number of indicators they are classed as eligible to join the
program. This system is simple, but it does not achieve the subtlety and
accuracy of results possible from a weighted points system. The main
drawback with the weighted points system is that it tends to give results that
give an appearance of greater accuracy than they can actually achieve. The
indicators are each general estimations of poverty level, not accurate
measures. Together they can triangulate each other and give a reasonable
estimation of poverty level. The scores are therefore best interpreted as
broad bands rather than accurate poverty scores.
Obviously the weighted point
system depends on a very thorough understanding of poverty, and the system can
only be as good as the relationship between each indicators and the poverty it
is measuring. Thus to develop a point system is time consuming and
difficult.
How can the system be monitored
and made rigorous?
The rigor and accuracy of the
check-list system depends on the understanding of poverty developed, and the
strength of the individual indicators chosen. Indicators can have a strong
link with poverty, but may also be subject to high occurrence of anomalies,
where the person is poorer or better off than the indicator may
suggest. Triangulation of a number of indicators will help to control for
this problem, but it is essential that:
1) there is an effective appeal system for dealing with people
who are wrongly classified
2) that there is on-going monitoring of the system to check
that the indicators maintain a strong link with poverty, and that the number of
anomalies does not become unacceptably high.
Inaccuracies may also occur in
the information given by the potential client or in observations made by the
loan officer. These types of errors can be controlled by including a mix
of indicators where information is given through interviews and those where
there is direct observation by a loan officer.
To develop an effective and rigorous
check-list tool requires considerable time and resources to research poverty,
and choose and test appropriate indicators. Check-list tools are
relatively time consuming, requiring detailed interviews with each potential
client – this takes up not just the interview time, but the time to travel to
each house, and for general greetings and conversation. It is therefore
impossible to use the tool to classify all of the members of a community.
Rather it can be used to screen applicants to the program to assess whether
they qualify to join, or simply to record the poverty level of clients.
For organizations such as Cashpor and SEF that seek to identify the poorest
people in a community and then motivate them to join the program a check-list
tool would be too time consuming and costly on its own. However, where
combined with another tool such as a the House Index, a check-list tool can be
very effective in accurately classifying those people already identified as
most likely to qualify for the program.
We have outlined three
poverty-targeting approaches that are reliable and cost-effective. The
choice of which tool to use depends largely on the objectives of using a
poverty-targeting tool, and in which context it is to be used.
Where there is a desire to understand local concepts of
poverty, to set up a transparent identification process, or where poverty
indicators are very variable, PWR may be the preferred tool. In contexts
where the sense of community is very weak, there are high levels of conflict,
or there are strong barriers to freely talk with women, PWR may be difficult to
implement. Similarly, if the skill levels of staff are low, then PWR may be
difficult to use.
For organizations simply wishing to measure the poverty
level of clients who join the program, or to screen out potential clients above
a certain wealth level, the check-list approach may well be
appropriate. This saves resources by not classifying the whole community,
and by focussing on actual or potential clients. The check-list approach
may also be useful as an impact measurement tool, since it can be applied to
any client at any time. However, for those organizations wishing to
actively recruit the poorest people in a community the check-list approach is
not sufficient on its own, and needs to be combined with a more general tool
such as the House Index. This obviously does raise the costs, and the two
stage process is likely to be more time consuming and costly than PWR.
The CHI relies on there
being a strong correlation between housing conditions and poverty. This is
not a universal relationship, and is very much defined by the
context. Where the CHI is adapted to local conditions, perhaps including other
externally visible, non-housing indicators, there is a greater chance of the
Index being applicable to a wider range of contexts. The CHI effectively combines a general screening of the
whole community, which identifies likely clients, with a more detailed
check-list tool – the NetWorth Test – that provides more detailed and accurate
poverty information. The House Index, and other visual targeting tools are
effective in contexts where there is a relatively uniform picture of poverty
and there is a strong relationship between a visible characteristic such as
housing and poverty.
Thus, in developing and adapting a poverty-targeting tool,
a number of choices need to be made that will determine which tool is
used. A critical point with any of the approaches is that the tools must
not be implemented blindly, but adapted to local conditions. In PWR, the
facilitation of the process will need to be developed to suit local
norms. In the CHI, there should be careful work to determine the effectiveness
of different visual indicators, and to develop a system that is based on the
best visual proxies, rather than assuming the housing structure is
effective. With the check-list tools considerable effort must be put into
developing appropriate poverty indicators.
Relating Poverty-Targeting Tools to National Poverty-Line
Measurements and to the Summit’s Goal of the Bottom 50% Below the Poverty Line
Neither the CHI, PWR nor the check-list tools reviewed
explicitly link their criteria to other “objective” poverty measures. Both are
able to define “poorest,” “poor,” and “non-poor” but these categories do not
automatically correlate with the income levels defined by the Summit’s
objectives. Although the CHI Net-Worth test and other check-list tools do
provide some more detailed information, PWR explicitly uses local definitions
of poverty, which are often not income based.
Where analysis of income level is a requirement for a
poverty tool, either method could be analyzed to provide the necessary data,
although this would have cost implications. The CHI (or an adapted visual
indicator test) provides a fixed list of characteristics. The Net-Worth
test or other check-list tool can then be used to sample qualifying households
at different levels against “poverty-line” economic data in order to
“calibrate” the index. This could be done in a similar way with PWR,
however, the subjectivity of results from each village could make
generalization difficult. Alternatively, economic data provided in the PWR
process (housing, education, food, income, expenditure, etc.) could be analyzed
and numeric values given, so as to approximate income levels. Proxy
indicators can also be used to equate the PWR results to economic
measurements. In the case of SEF, for example, the position of state
pensioners in the ranking, who receive a known income from the government, has
proved to be an effective proxy that makes comparison of the PWR to national
poverty-line figures possible.
In this paper we have demonstrated the central role
poverty-targeting must play in any poverty-focused microfinance institution,
not just in terms of knowing and reporting on who we are reaching, but as a
fundamental basis for building a sustainable program, which is designed around
and which meets the needs of the poorest.
Reliable and cost-effective poverty-targeting is being
achieved in practice, and has been operationalized into the day-to-day
operations of many MFIs. The obstacles that, in the past, led MFIs to
avoid targeting and create programs that did not reach the poorest, as well as
claim they could not report on who they were reaching, have now been overcome.
SEF has been using PWR as the starting point in new
villages for about one and a half years. During this period, ranking has
taken place for over 20,000 households. SEF targets the poorest people as
defined by communities themselves, but has demonstrated that this approximates
to the Summit’s goal of the bottom 50% of households living below the poverty
line. There is obviously some leakage, but this is mostly at the
borderline between the poorest and the poor. Field staff and community
members report that the program is effectively reaching the poorest, and that
the vast majority of clients are in SEF’s target group.
The CHI has been implemented widely in Asia and the
Pacific, and a recent evaluation concluded that 97% of clients are within their
target group of the poor and very poor.
It is hoped that this paper, and the methods outlined, will
bring a new awareness, understanding, and commitment within the international
development community to reaching the poorest families. We urge practitioners
to take up the challenge of reaching the poorest, and adapting and implementing
targeting methods that make sense in their own contexts.
ANNEX 1. SOME EXAMPLES OF CHECK-LIST
TOOLS
Means
Test: Kabalikat para sa Maunlad na Buhay, Inc. (KMBI/Philippines)
KMBI in the Philippines has created a composite poverty
assessment instrument it calls the Means Test Form. This single-page household
interview form consists of: (1) an unscored section for borrower background
information (address, age, education, civil status, business experience), (2)
an unscored section on income (current sources for spouse and immediate family
members), (3) a 5-variable housing index, (4) a 12-variable asset index, and
(5) a box for estimating the composite score. The scoring system—based only on
the sum of the housing and asset indices—creates five levels of poverty ranging
from level 1 or poorest (4-15 points) to level 5 or wealthy (46-55 points).
KMBI excludes potential clients with scores higher than 30. It is noteworthy
that both the housing and asset sections do not require numerical estimates by
the potential client; rather, the interviewer simply checks boxes with
predetermined scores. Furthermore, the reported income of the household is a
reference point only and does not affect the final score. Finally, data on
household liabilities are not requested.
The means test is fairly simple, because even though it
asks more than ten questions it is restricted to one page and is more succinct
than similar tools used. This simplicity means that it can be built in as a
routine function of field staff screening of clients. Its cost is fairly
similar to that of the CHI and PWR, requiring less than 40 minutes per client
to complete. Its point system does a good job of discriminating among the very
poor, poor, and nonpoor. It also produces high quality data because of the
precoded checklist system of answers, which has built-in cross checks, although
the process is open to abuse, and requires an effective system of spot-checking
by a supervisor.
It is interesting to note that KMBI also uses the
information collected in the Means Test as a baseline for subsequent impact
evaluations.
RHUNU
UNESCO, Mr. C.A. Samaradivakara
At
Rhunu Unesco in Sri Lanka, a card is completed for each beneficiary family with
information gathered from a survey done of the family’s
circumstances. Points are determined for certain levels within each
category, although unlike other tools, there is no weighting for the points
from different indicators. Points (0-10) are awarded for each indicator; the lower
the overall score, the poorer the family. Families scoring 30 points or
less are considered living below the poverty line in Sri Lanka, whereas
families scoring 20 points or less are considered in the bottom half of those
living below the poverty line. The survey includes:
1)
Monthly income for a family: (US$50 = 10 points / $40 = 8 points / $30 = 6
points / $20 = 4 points / $10 = 2 points)
2)
Quality of housing: (Permanent = 10 points / Semi-permanent = 5 points /
Temporary = 0 points)
3)
Health:
a.Access
to pure water within 100 yards (Owns or has access to pure water source = 2.5
points / No access to pure water source = 0 points)
b.Access to proper toilet
facilities within 100 yards (Permanent water-sealed toilet = 2.5 points /
Water-sealed toilet needing improvement = 1 point / No toilet facilities = 0
points)
c.Nutritional
level of the children (as evidenced by a Child Development card maintained by
the government = up to 2.5 points)
d.Immunization
of children (as evidenced by a Child Development card maintained by the
government = up to 2.5 points)
4)
Number of school-going children: (Full score = 10 points / 2 points deducted
for every child ages 6 - 18 that is not attending school)
4)
Availability: (Electricity =
10 points / No electricity = 0 points)
FAMILY DEVELOPMENT FUND, UNICEF
The
Family Development Fund (FDF) in Egypt uses a tool that targets some of the
poorest villages of Upper Egypt, and their target population is very poor
women. Female high school graduates from the community are selected as
loan officers and trained to do assessments. The tool is very simple,
using a small number of indicators and no point system. This would be very
weak on its own, but FDF combines the check-list tool with a final verification
by a community-based loan committee. Information is not available for the
functioning of these committees, but if they are effective then this would be a
very low cost tool.
Criteria
for selection include the following:
2) Per
capita income of the family members is not more than Egyptian Pound 40.00/US$12
per month (based on interviews with the women regarding their sources and
amount of income)
3) Land
owned or leased should not be more than 4 Kerats (total 0.17 acres)
4)
Women eligible for the Sadat Pension Scheme and receive social security from
the Ministry of Social Affairs
Priority for loans is given to
those who meet more than three items of the above criteria.
INTERNATIONAL
RESCUE COMMITTEE SEAD PROGRAM, Sunimal Alles
The
International Rescue Committee's SEAD program in the Ivory Coast works
primarily with Liberian refugees in the Ivory Coast. Staff carry out
baseline surveys and document findings before any assistance is
provided. The loan staff observe for themselves, as well as interview
participants and their neighbors, concerning several items differentiated by
the titles very poor, not so poor, non poor.
The
tool is simple and has well thought out indicators which relate well to
local poverty conditions. It uses proxies for income and expenditure, rather
than trying to get direct answers - all the indicators can be either observed
or information gathered with a simple question, and this should not therefore
be a time-consuming process. There is also a good mixture of proxies which can
be directly observed and those that are asked - this helps to save time, and
also ensures triangulation of information.
Although
the tool achieves simplicity, it is not very effective in accurately
distinguishing between different levels of poverty, using only three
categories: Very Poor (VP), Not So Poor (NSP), Non Poor (NP)
1)
Cooking Utensils: (VP = Leaking, damaged, limited quantity / NSP = Better pots,
1 set of dishes / NP = Better quality, more than 1 set)
2)
Sleeping condition: (VP = Floor or mat / NSP = Mattress / NP = Bed)
3)
Clothes: (VP = Torn or dirty / NSP = Patched / NP = Good clothes)
4)
Slippers: (VP = No footwear, different models, damaged / NSP = Repaired and
second hand / NP = Better/new shoes and slippers)
5)
Food: (VP = Dry rice, 1 meal a day / NSP = Rice with soup, 2 meals a day / NP =
Rice with meat/fish, 3 meals a day)
6)
Health condition: (VP = Skin rash, infected eyes, sore feet, cough, running
nose, diarrhea / NSP = Some of the same as for the very poor, but better
health, buys medicines from street vendors / NP = Visits to the doctor,
medicines from pharmacy)
7) Schooling:
(VP = No children in school / NSP = Boys in school / NP = All children in
school)
8)
Housing: (VP = Leaking house, no door, cloth to cover entrance / NSP = 1 front
door, 1 room cemented / NP = Cemented house floor, 2 doors, good roof, 4
windows, flowers in yard)
9)
Furniture: (VP = No chairs, some benches / NSP = 2 chairs or stools / NP =
Sufficient furniture)
10)
Utilities: (VP = No toilet, 1 or no lantern, uses fire for light / NSP =
Shallow pit latrine, 2 lanterns, well for water / NP = Covered well for water
pump, flushing toilet)
11)
Domestic employees: (VP = none / NSP = none / NP = 1 employee)
12)
Transportation: (VP = Walking / NSP = Paying for taxi / NP = Bicycle or
motorbike)
13)
Radio: (VP = None / NSP = Old or damaged / NP = Good or new)
14) Ice
box: (VP = None / NSP = Clay pot to cool water and drinks / NP = Refrigerator)
15)
Animals: (VP = None / NSP = Sheep, chickens / NP = Cows, sheep)
16)
Business income: (VP = No income or working as laborer / NSP = Part time / NP
=Regular employment)
17) Economic activities: (VP =
Gathering and selling palm nuts and wood / NSP = Buying and selling wood, table
market / NP = Larger table market, store, cosmetic shop)
LIFT
ABOVE POVERTY ORGANIZATION (LAPO), Uwa Izekor
The
Lift Above Poverty Organization in Nigeria developed their poverty measurement
tool after visiting Grameen Bank in 1990.Their tool provides a scoring system
between 25 and 100, with the higher scores indicating greater poverty. People
are eligible for a loan from LAPO if they score 50 points or above, as their
economic situation would correspond similarly with people living below
Nigeria's official poverty line. LAPO staff have indicated that people
scoring 70 points or above would have economic situations that correspond
similarly with the Summit's target group, the bottom half of the population
living below the Nigerian poverty line. The questions take five minutes to
complete per client. This is used as an initial cut-off to determine
eligibility. Before a client receives a loan, there are five more meetings
with loan officers who can use these meetings to verify the accuracy of the
information provided in the first interview.
The
LAPO tool is similar in some respects to that of the IRC, however it
importantly does not attempt to ask actual figures for monthly income and
business worth. The main strength of this tool is that the indicators are
weighted, however, it is not clear whether the indicators are rigorous enough to
be used effectively in this type of weighted system.
The
list of criteria include:
1)
Personal:
a.
Level of formal education:
(None = 12 points / Half primary = 8 points / Full primary = 6
points
/ Half secondary = 4 points / Full secondary = 2 points)
b. Number of dependants under 20 years of age: (Above 9 =
10 points / Between 6 and 9 = 8
points / Between 3 and 5 = 5 points / Between 1 and 2 = 3
points / None = 2 points)
c. Marital status: (Widow = 10 points / Separated or
divorced = 6 points / Married = 5 points /
Single = 4 points)
2) Household:
d. Building: (Rented = 12 points / Inherited = 6 points /
Own = 3 points)
e. Size of dwelling place: (1 room = 7 points / 2 rooms = 5
points / 3 rooms = 3 points /
Flat/Bungalow = 2 points)
f.
Highest form of entertainment owned: (None = 12 points / Radio = 6 points /
T.V. = 1 point)
3)
Business/Occupation
g1. Worth of business [in Naira US$1 = 88 NGN]: (No
business/1000-5000 = 15 points / 6000
10000 = 10 points / 11000-20000 = 8 points / 21000-50000 =
5 points / 5100 and above = 3
points)
OR
g2.
Size of farm: (Small = 15 points / Medium / 9 points / Large = 3 points)
h. Monthly income: (Below national minimum wage = 12 points
/ Above national minimum
wage = 5 points)
4) Location:
i. Location: (Rural = 10 points / Semi-urban or Local
Government Headquarters = 6 points /
Urban or State capital = 3 points)
SCORE SHEET (Maximum,
Minimum)
1) Level of formal education (12
points, 2 points)
2) Number of dependants under 20 years of age (10
points, 2 points)
3) Marital status (10
points, 4 points)
4) Building - house (12
points, 3 points)
5) Size of dwelling place (
7 points, 2 points)
6) Highest form of entertainment owned (12
points, 1 points)
7) Worth or business OR Size of farm (15
points, 3 points)
8) Monthly income (12
points, 5 points)
9) Location (10
points, 3 points)
TOTAL (Maximum = 100 points / Minimum = 25 points)
ANNEX
2.CASE-STUDY: PWR IN BUNGENI, SOUTH AFRICA
Characteristics of Different Wealth Groups in Bungeni
Poorest:
*
single parent unemployed, or two parents both unemployed
* many children
* being unmarried and having no family to assist
* dependent on temporary jobs
* no means of provision except by begging
* widows with many children
* insufficient and poor quality food - often have to beg food
* no proper place to sleep - poor quality housing
* orphans with no parents
* inability to educate children
* few clothes - almost never buy
* no assets
* no self-respect or respect from others
Poor:
*
temporary jobs (e.g., farm laborers)
* have some food, but struggle
* working widows and pensioners with many children
* parents dependent on working children who also have their
own families in the samehousehold sharing resources
* working on agriculture scheme
* many children
* no pension/pensioners with many children
* unmarried
* have some house (though not good) - some made of mud bricks, with cracks
* can provide something from their temporary jobs
*
children attend school irregularly
Quite poor:
*
earns enough to cope daily - mostly temporary work/ self-employed
* those with smaller number of children to look after
* pensioners with less children
* widows with pensions from late husbands
* have sleeping place
* unmarried
* pay-outs from old jobs
* children complete primary school
* able to buy enough food
OK:
*
pensioners with only themselves to look after
* few children
* good supply of food - varied diet
* families where at least one parent has a permanent job
* children attend school regularly
* good house
Wealthy:
*
professionals and business owners
* good money to
adequately supply their family
* children attend school
properly
* electricity in the
house
* own a television
* smaller families
* own a car/gun
* eat bread with
margarine
* children nicely clothed
* children attending
tertiary education
* company pensions
* food in abundance
* excellent housing
* drink tea every day
REFERENCES
Chambers, Robert (1997); Whose Reality
Counts?Putting the First Last; Intermediate Technology Publications, UK
de Wit, John (1997); Presentation to the Microcredit Summit
on poverty targeting; the Small Enterprise Foundation, South Africa
Gibbons, D.S. and Meehan, Jennifer (1999); The
Microcredit Summit’s Challenge: Working Towards Institutional Financial
Self-Sufficiency while Maintaining a Commitment to Serving the Poorest Families,
Paper presented at the 1999 Meeting of Councils of the Microcredit Summit
Campaign, Abidjan, Cote D’Ivoire
Gibbons, D.S. and Tomlinson, W. ,Eds (1995); Draft Training
Manual On Cost-effective Targeting, For Credit And Savings Program For The Poor
Basic Series In Management Development Training, produced by CASHPOR and
published By Grameen Trust
Gottschalk, Jan (undated); “The Visual Indicator of Poverty
and the Poverty Test”; Working Paper for the Small Enterprise Foundation,
Tzaneen, South Africa
Grandin, Barbara E(1988); Wealth Ranking in Smallholder
Communities: A Field Manual; Intermediate Technology Publications,
UK
Hatch, John and Frederick, Laura (1998); Poverty
Assessment by Microfinance Institutions: A Review of Current Practice; Finca
International, paper for Microenterprise Best Practices project of USAID
Hulme, D. and Mosely, P. (1996); Finance Against Poverty
Volume 1 & 2, London & New York, Routledge
Johnson, Susan and Rogaly, Ben (1998); Microfinance and
Poverty Reduction; Oxfam, Oxford, UK
Johnson, Susan (1997) “Gender and Microfinance: guidelines
for good practice”; paper produced for ActionAid.
Mayoux, Linda (1998); “Microfinance and Women’s
Empowerment: Approaches, Evidence and Ways Foward”; Open University Development
Policy and Practice Discussion Paper No. 41, Milton Keynes, UK
Maxwell, Simon (1999); “The Meaning and Measurement of
Poverty”; ODI Poverty Briefing, Overseas Development Institute, London, UK
(available on http://www.oneworld.org/odi/briefing/pov3/html)
SALDRU (1994); South African’s Rich and Poor: Baseline
Household Statistics; Project for Statistics on Living Standards and
Development, University of Stellenbosch
Todd, H and Simanowitz, A, with Nkuna, N (1999);“A Manual
for Poverty-targeting in Microfinance: Participatory Wealth Ranking and the
CASHPOR House Index”; CASHPOR Inc., Malaysia, Publication forthcoming
Simanowitz, A (2000);“Targeting the Poor: A Comparison
between Participatory and Visual Methods”;Small Enterprise Development: an
International Journal; Vol. 11, No. 1, March 2000
Todd, Helen (1996): Women At The Center: Grameen Bank
Borrowers After One Decade, West-view Press
Anton Simanowitz and Ben Nkuna’s work in
contributing to this paper was in part made possible through support provided
by The Ford Foundation and the Office of Microenterprise Development, Economic
Growth and Agriculture Development Centre of the Global Bureau, U.S. Agency for
International Development, under the terms of Award No.
PCE-A-00-98-00039-00. Opinions expressed herein are those of the authors
and do not necessarily reflect the view of The Ford Foundation or the U. S.
Agency for International Development. The support of these agencies is
acknowledged and we thank them for their assistance in enabling us to carry out
and to publicize action research into the elimination of poverty.