Sie sind auf Seite 1von 4

ASSESSING THE RELATIVE POVERTY OF MICROFINANCE CLIENTS:

A CGAP OPERATIONAL TOOL

The CGAP Poverty Assessment Tool provides transparency on the depth of poverty outreach of MFIs. It provides rigor-
ous data on the levels of poverty of clients relative to people within the same community through the construction of a
multidimensional poverty index that allows for comparisons between MFIs and across countries. It has been primarily
designed for donors and investors who would require a more standardized, globally applicable and rigorous set of
indicators to make poverty-focused funding decisions. The tool involves a survey of 200 randomly selected clients and
300 non clients, takes about four months to complete and costs around $10,000. Field tests were successfully complet-
ed in four countries. The Poverty Assessment Tool should be used in conjunction with other appraisal tools (such as the
The Consultative
CGAP Appraisal Format) to ensure a holistic understanding of MFIs.
Group to Assist
the Poorest The Context Development of the Tool
CGAP is committed to the twin objectives of increas- CGAP commissioned the International Food Policy
1818 H Street, NW ing the financial and institutional sustainability of Research Institute (IFPRI) to work with CGAP staff
Washington, DC
20433
microfinance institutions (MFIs) and deepening their (Brigit Helms, Syed Hashemi and Elizabeth Littlefield)
poverty focus (i.e. increasing poverty outreach and to develop a simple, operational tool that would provide
Tel: 202 473 9594 impact on poorer people). As part of this commitment, reliable statistical information on the poverty level of MFI
Fax: 202 522 3744
and in its role as a service provider to the microfinance clients at the lowest possible cost. The IFPRI team, led
Email: industry, CGAP has continually endeavored to pro- by Manfred Zeller, conducted extensive literature review
cgap@worldbank.org vide tools that allow for greater transparency of MFI and expert consultation on the general use and applica-
Web:
performance in meeting such objectives. The CGAP bility of poverty indicators. They then developed a set
www.cgap.org Appraisal Format was developed to provide practical of indicators reflecting the multidimensionality of pover-
guidelines and indicators for measuring MFI per- ty (income potential, asset ownership, human capital,
formance in governance, management and leadership, social capital, nutritional levels, access to food, shelter and
mission and plans, systems, operations, human resource clothing, security, etc). The indicators were selected on
management, products, portfolio quality and financial the basis of their reliability, simplicity, discriminating qual-
analysis. Analysis of these institutional features would ity and universal applicability. A generic questionnaire,
in turn allow for an appraisal of the potential for insti- open to changes and revisions given different social and
tutional viability and sustainability. cultural contexts, was also prepared. This questionnaire
The focus on transparency thus far, has centered was then used with local research firms in surveys of clients
around financial performance. There has been a marked of four MFIs in four different contexts in differing geo-
under-emphasis on the equally important issue of trans- graphical areas (two in Africa, one in South Asia, and one
parency of the poverty focus of MFIs. The CGAP Pover- in Central America). The findings of the surveys, the con-
ty Assessment Tool has therefore been developed as a much struction of poverty indices, and the resulting measure-
needed tool to provide transparency on the depth of ment of MFI depth of outreach were shared with the
poverty outreach of MFIs. Using the tool involves ran- MFIs, grassroots practitioners, academics and other peo-
dom sample surveys of clients and non-clients. Statisti- ple in the microfinance industry. Inputs from such dia-
cal analysis provides rigorous data on the levels of poverty logues were used to revise the manual. The entire process
of clients relative to people within the same communi- of developing the tool took a year and a half, and seven
ty. The tool constructs a multidimensional poverty index months of CGAP staff time.
that allows for comparisons between MFIs and across
countries. Targeting tools (such as the housing index, Who Is the Tool For?
means testing and participatory wealth ranking) have The tool is intended primarily for use by donors that
long been used to effectively target services to very poor focus on poverty outreach to form part of their apprais-
people. The Poverty Assessment Tool is not a substitute al of MFIs for funding. It should be used in conjunc-
for these. The Poverty Assessment Tool is far more com- tion with other appraisal tools (such as the CGAP
plicated, time-consuming and costly. It has been prima- Appraisal Format) to ensure a holistic understanding
rily designed for donors and investors who would require of MFIs. While the tool is relatively simple it still pre-
a more standardized, globally applicable and rigorous supposes a certain level of statistical skills and famil-
set of indicators than what targeting tools could provide. iarity with the statistical package SPSS. It is expected
therefore that the team implementing the tool have Housing indicators: ownership status, room size, build-
both a social sciences background with knowledge of ing material, access to electricity, drinking water and
poverty issues as well as basic statistical experience. sanitation, cooking fuel
Donors would hire such teams locally. Our experience Land ownership
has been that most countries have research institutions Ownership of assets: livestock, productive assets and
and private firms that are well qualified for such activ- consumption assets
ity. In all the field tests concluded so far, CGAP has The emphasis on a variety of indicators stem from
used local teams. Given the need for technical expert- a realization that poverty is multidimensional and that
ise and the resources involved in running such surveys, any one indicator would not be capable of capturing
MFIs themselves are generally not encouraged to take poverty levels across different countries and contexts.
on the tasks themselves. This is primarily a donor tool For example while land ownership in rural Bangladesh
to inform them of the depth of outreach of MFIs, to and housing structure in South and South East Asia are
better make funding decisions. Of course the infor- very good indicators of economic levels there, they
mation does have great value for MFIs. have very little relevance for parts of Africa and Latin
America. Additionally the use of multiple indicators pro-
How Expensive Is It to Use the Tool? vides for a better approximation of poverty levels.
The cost of surveys that CGAP has conducted range
from $4,000 to $16,000. The average cost for con- Data Analysis and Construction
ducting the survey should be around $10,000. The of the Poverty Index
assessment of poverty levels of MFI clients, from The survey manual explains the data analysis process
adapting the questionnaire, to conducting the survey, using SPSS, probably the most widely used software
analyzing the data and preparing the final report, package. Of course any other statistical package can
should take about four months. be used but the construction of the Poverty Index
using principal component analysis is best done with
Sampling Design SPSS. While the analysis is not overly complicated it
A sample size of 200 MFI clients and 300 comparison is assumed that analysts will have some competencies
households is recommended to ensure a balance between in statistical analysis.
reliability and the high costs involved in much larger sur- Simple crosstabs will provide immediate compar-
veys. The household is the basic sampling unit. Only isons of clients and non-clients in terms of different
new MFI clients (less than six months into the program) indicators. However the key feature of the Poverty
are sampled to eliminate any impact from being in the Assessment Tool is the Poverty Index. The Poverty
program. Comparison households are selected ran- Index is constructed through the application of prin-
domly from the same geographical area of MFI opera- cipal component analysis (PCA). The PCA method
tion so that comparisons can be made between those that is applied to determine how information from vari-
are selected to be in the MFI program and those who ous indicators can be most effectively combined to
represent the general community. measure a households relative poverty status. Which
combinations of indicators prove the most instru-
Dimensions of Poverty mental in measuring relative poverty in a given sur-
The survey tries to elicit information on different dimen- vey area will differ, and often in ways that are somewhat
sions of poverty. However the questionnaire has been
designed to be simple and operational. Obviously, spe-
cific questions and the wording of each question has to Figure 1. Determining cut-off scores for
be modified and adapted in each study to ensure that poverty ranking
it addresses the social and cultural contexts of the region
where the study takes place. For example to determine Client households Client households Client households
with scores with scores between with scores
levels of food consumption, information is collected on less than .70 .70 and 0.21 above 0.21
the consumption of inferior food. However the spe-
cific inferior food used in the questionnaire varies from Lowest Middle Higher
cassava in Kenya to coarse bread and chili in India to
Poverty Score Index
tortillas in Nicaragua. Similarly for asset ownership dif-
ferent commodities would be used in different coun- 2.51 0.70 0.21 3.75
tries representing the best indicators for marking different
economic levels of households. Bottom 100 Middle 100 Top 100
The survey collects information on households on non-client non-client non-client
the following dimensions: households households households

Demographic structure and economic activities


Footwear and clothing expenditure
Food security and vulnerability: frequency of meals, con- Cutoff scores
sumption of luxury and inferior food, hunger episodes
predictable. In countries where poverty is extreme, the client and the non-client population. For instance,
indicators signaling chronic hunger tend to differen- if 60 percent of the client households fall into the first
tiate the relative poverty of households. In densely pop- tercile or lowest poverty category, the MFI reaches a
ulated countries, ownership of land and dwellings disproportionate number of very poor clients relative
may better signal differences in relative poverty. The to the general population.
end result of PCA is the creation of a single index of Figure 2 shows the results of one of the four case
relative poverty that assigns to each sample household studies highlighting significant differences in the pover-
a specific value, called a score, representing that house- ty distribution between clients and non-clients. The
holds poverty status in relation to all other households graph shows that clients are over-represented within
in the sample. The lower the score, the poorer the the lowest tercile and under-represented in the high-
household relative to all others with higher scores. The est tercile. This would indicate that the MFI is reach-
scores of MFI client households and non-client house- ing a larger share of poorest households than what is
holds are then compared to indicate the extent to found in the population in general. In contrast, the
which the MFI reaches the poor. results of another case study found the opposite pat-
First, however, the share of the local population that tern. In Figure 3, the results indicate that MFI clients
is likely to fit the assessments definition of poor must are under-represented in the lowest tercile and over-
be decided on. CGAP, in its assessments, has used a represented in the highest tercile.
cutoff of 33 percent of the control population to
define the poorest group within the local popula-
tion. This decision is based on the usefulness of cat- Figure 2. MFI with extensive poverty outreach
egorizing local populations into terciles that can be
broadly interpreted to represent the lowest, middle and
Percent
higher ranked groups of households ranked by rela- 60 MFI client
tive poverty. The methodology can be adapted to Non-client
include additional categorization. If half of the peo- 50
ple in a country are below the poverty line and if the
bottom 25 percent are said to be the hard-core poor, 40
then the local population (the comparison group)
can be divided into quartiles. Other divisions can 30
also easily be made.
Each assessment study includes a random sample 20
of 300 non-client households and 200 client house-
holds. To use the Poverty Index for making compar- 10
isons, the non-client sample is first sorted in an
ascending order according to its index score. Once sort- 0
Lowest Middle Higher
ed, non-client households are divided in terciles based
on their Poverty Index score: the top third of the non- Poverty group
client households are grouped in the higher ranked
group, followed by the middle ranked group and
finally the bottom third in the lowest ranked group.
Since there are 300 non-clients, each group contains Figure 3. MFI with lower poverty outreach
100 households each. The cutoff scores for each ter-
cile define the limits of each poverty group. Client Percent
households are then categorized into the three groups 60
MFI client
based on their household scores. Figure 1 illustrates the Non-client
use of cutoff scores to create poverty terciles from 50
non-client households. The cutoff scores of .70 and
+. 21 were calculated from an actual case study exam- 40
ple. Each poverty assessment will use different cutoff
scores to group households. 30
Now that all cases for MFI clients and non-clients
have been assigned to poverty groupings, comparing 20
differences between the two distributions is possible.
If the pattern of client households poverty matches 10
that of the non-client households, client households
would divide equally among the three poverty group- 0
Lowest Middle Higher
ings just as the non-client households, with 33 per-
Poverty group
cent falling in each group. Hence any deviation from
this equal proportion signals a difference between
Integrating a National Comparison this index lies in it being simple, observable and ver-
The Poverty Index allows for a comparison of pover- ifiable. Since dwellings often represent the largest
ty levels of clients and non-clients in a specific region. investment of a household and since in many regions
It does not say, however, whether the region itself is there are specific variations in housing patterns
in a better-off or worse-off area within the country or (makeshift hut, to thatched house, to tin roof house
even whether the MFI is situated in a poor or wealthy to brick and concrete houses) reflecting differences in
country. The CGAP tool therefore supplements the economic levels, housing can be used as an excellent
client survey information with two additional inputs. proxy for ranking households. The problem, howev-

www.cgap.org
The first is a national evaluation by a panel of experts er, is in generalizing such an indicator across rural
who rate the MFIs operational area against a nation- and urban differences and across countries. Slum
al average. The rating is based on assessments of wages dwellers often live in brick and concrete houses but in
and employment levels, physical and social infra- far worse conditions than rural families in thatched or
structure, literacy levels and agricultural conditions (if tin houses. Other indicators, such as land ownership,
rural). The second is the use of the Human Develop- work well as proxy indicators in specific contexts but
ment Index to rank the poverty level of the country cannot be generalized as valid across regions.
against all other countries. These then allow apprais-
ers to determine that even if MFI clients may not be Use of the CGAP Poverty Assessment Tool
very poor relative to non-clients, they still may be The Poverty Assessment Tool is very clearly not a tar-
reaching very poor people if, for example, they are sit- geting tool to be used by MFIs. The Housing Index,
uated in an economically depressed area in one of the the Participatory Wealth Ranking exercises, Means
poorest countries in the world. Tests and other tools currently being used generally pro-
vide good, low-cost targeting information for MFIs.
Comparisons with Alternative Poverty Since the CGAP Tool is statistically rigorous and far
Assessment Approaches more accurate, comparisons of such targeting method-
There are several methods to determine poverty lev- ologies with the CGAP Tool at some point in an
els of populations that are used by academics, policy MFIs growth, however, would indicate the effective-
makers and development practitioners. ness of the targeting strategy being used.
Detailed Household Expenditure Surveys including The Poverty Assessment Tool is not a market research
the Living Standard Measurement Survey (conduct- tool though information from it could be used by
ed by the World Bank) are used nationally for pover- MFI management to understand economic levels of
ty analysis. These surveys are extremely rigorous and their clients. It is also not a baseline survey for use by
provide for more accurate information than CGAPs MFIs though the questionnaire can be adapted and
Poverty Assessment Tool. However such surveys use used as part of a baseline design for impact assessment
large samples and detailed data collection that make and impact monitoring.
it far too costly, time consuming, cumbersome and ana- The Poverty Assessment Tool provides rigorous cri-
lytically too demanding for use on regular MFI teria with which to determine the depth of MFI out-
appraisals. reach. It forces donors to integrate a poverty focus in
Rapid Appraisals and Participatory Appraisals are their appraisal of MFIs and in their funding decisions.
often used to generate fast information on local level CGAP strongly believes that the future of the microfi-
economic conditions. Rapid Appraisals using key nance industry lies in moving beyond the poverty-sus-
informants can often be subjective and even those tainability polemic in favor of pushing the microfinance
using discussions with larger community groups leave frontier forward on both poverty outreach and sus-
open problems of reliability. Participatory Appraisals, tainability. There is great scope to creatively improve on
including Participatory Wealth Ranking1 used by some both without sacrificing the other. The Poverty Assess-
MFIs, offer a far more reliable method for communities ment Tool, by making the depth of outreach more
themselves to identify who the poor are. This allows transparent, forms an important step to guide us to more
for a more holistic and people-centric determination effective supporting of a broader range of MFIs.
of poverty and of ranking. A probably more impor-
tant function of such methodology is the empower- Note
ment of the community. It asserts the primacy of local 1. For more information on both PWR and HI,
knowledge over externally determined measurement refer to the excellent piece by Anton Simanowitz, Ben
criteria and lets the community take charge in decid- Nkuna and Sukor Kasim Overcoming the Obsta-
ing how rankings are to take place. However while this cles of Identifying the Poorest Families: Using Par-
approach works well in identifying the poor in each ticipatory Wealth Ranking (PWR), the CASHPOR
village or neighborhood it cannot be used to rank House Index (CHI), and Other Measurements to
larger populations or determine the poor(est) in a Identify and Encourage the Participation of the Poor-
large geographical region. est Families, Especially the Women of Those Fami-
Many MFI practitioners use the Housing Index to lies, commissioned by the Microcredit Summit,
determine poverty levels of households. The appeal of June 2000.

Das könnte Ihnen auch gefallen