|
|
 |
Participation
in Micro-credit Programs and Household Social Well-being
6.1
Background and Hypotheses
It is now well accepted
that, within the process of socio-economic development, economic
growth (rise in income/consumption levels) and the development
of human capabilities (health, education, skills, knowledge) are
synergistically related. This is because with increase in incomes,
households tend to invest more in raising the human capabilities
of household members. Alternately, enhancement of human capabilities
has a positive effect on household’s income-earning capacity and
consumption levels. Hence, rise in household incomes and increased
employment (economic well-being) are generally accompanied by
improvement in the aggregate levels of social well-being. In fact,
reduction in household poverty is now measured not only in terms
of increase in income, but also in terms of gains in human capabilities.
However, because household
allocative decisions are often not according to individual need,
there tends to emerge a gap between requirement (need) and actual
consumption, even under situations of growth in household income
levels. Such gaps between consumption (of household resources)
and need (of individual family members) are manifested in terms
of systematic differences in outcomes by age and gender (for example,
in nutritional status and anthropometry, health status, schooling
and literacy levels, and other outcomes that reflect consumption).
Eventually, these differences get manifested in the broader societal
age/gender inequalities in well-being outcomes. It is widely recognized
that such differences within households and societies, on the
basis of age and gender, are not desirable for the long-term development
prospects of the society. Hence one observes the strong emphasis
in all of the international development targets on the removal
of all types of inequalities in outcomes.
Systematic age/gender differences
in outcomes are observed across socio-economic class, but may
become accentuated under certain conditions, such as household
poverty. This raises the question of the underlying causes for
the existing allocative patterns, or in other words, the explanation
for the observed pattern of outcomes. The underlying “economic”
rationale for the pattern of allocation is that households allocate
resources to ensure the immediate and long-term welfare of the
household, i.e, to maximize aggregate household utility. In other
words, under conditions of scarce resources, it makes economic
sense to allocate relatively more resources (food, health care,
schooling) to ensure the welfare of the earning members even at
the welfare cost of the non-earning members. Under this theory,
it is assumed that age/gender differences in outcomes will disappear
with reduction in the scarcity of household resources or with
rise in household income. But even if differences in well-being
outcomes are reduced as household income rises, they do not disappear,
suggesting that allocative patterns are subject to other non-economic
forces.
There are in fact other
compelling reasons for differential allocation of household resources
that are social/cultural in origin, and may be termed “irrational”,
since these emerge from discriminatory norms and practices that
are socially determined. These norms and practices are based upon
institutions and structures that determine legitimate behaviour,
as well as upon inadequate and inaccurate knowledge and information
about cause and effect. Hence, the removal of age/gender biases
in outcomes requires that both the scarcity aspect of household
resource allocation as well as the discrimination and knowledge
aspects must be addressed.
Most mainstream poverty
alleviation programs aim at increasing household income and at
raising the aggregate level of household well-being. They do not
address the non-economic reasons for outcome differences by age
and gender, especially the intra-household distributional aspect.
Within the general context of development, micro-credit programs
are seen as special development interventions aimed at bringing
about desirable change in both economic and social well-being
at the household level. There are two distinct dimensions of this
change: the improvement in aggregate household well-being and
the reduction in inequality and gender biases in well-being. Micro-credit programs help
to mitigate the scarcity of household resources by reducing the
credit constraint faced by households. This allows households
to raise income from self employment and increase consumption
levels. In addition, because of their tremendous capacity to reach
poor women, these programs are able to address intra-household
gender relationships and the distributional and knowledge aspects
of household resource allocation. Hence, micro-credit programs
are in a position to tackle both the economic and the non-economic
factors affecting household well-being, and to lead to improvement
not only in aggregate household well-being, but also to the reduction
of intra-household inequalities in well-being. Based on the above
assumptions, the following working hypotheses about the impact
of micro-credit program participation are suggested:
1.Participation
in MC programs improves aggregate household well-being, both social
and economic, and reduces household vulnerability.
2.Participation
in MC programs reduces household gender inequalities in well-being
and improves the position of
women within the household and in the community.
However, there
are several pitfalls in the measurement of the effects of micro-credit
program participation. The first is the inability to separate
the credit effects of program participation from the non-credit
effects, which are attributable to the social mobilization efforts
of programs and other non-credit inputs (such as social awareness,
literacy, leadership training, legal awareness, etc.) that accompany
the provision of credit. Second, the secular effects of other
processes in the broader economy and society, including other
development interventions, are difficult to isolate because of
the difficulty of generating data from appropriately designed
samples. Third, program participation suffers from selectivity
biases that distort the magnitude of observed program effects,
unless the effects of unobserved
factors are accounted for in the measurement of program
effects. The above measurement problems render program non-program
comparisons difficult in cross-sectional surveys, using the type
of simple experimental design conventionally adhered to in a survey.
Panel household data that compares change in outcomes over time
for the same household can overcome some of these measurement
errors, including those arising from selectivity biases. However,
this means that the aggregate or cross-sectional picture of change
over time is lost since the panel of households in the later survey
rounds is no longer representative of the population covered.
Moreover, since panel comparisons require the same households
to be surveyed, the numbers of available households in the different
cells can be quite small in the case of variables that are not
applicable for all households, such as access to schooling or
access to modern health care when ill.
6.2
Methodology
In order to assess the effect
of micro-credit program participation, the sample households are
categorized into three groups according to participation status:
continuous participants or households that were participants throughout
the study period from later 1997 to early 2000; occasional participants
or households that were participants during part of this period
or were on/off participants; and never participants or households
that never participated either before during the survey period.
Occasional participants include complete and recent dropouts as
well as new participants who joined after the first round survey.
These categories will allow comparisons among participating households
and between participation and non-participation. In other words,
the effect of participation per se as well as the effect of differential
participation with respect to regularity of participation can
both be examined.
The use of panel data allows the assessment of the independent
effect of program participation on social well-being. This is
because the effects of other non-program factors, including the
effects of unobserved selectivity factors, on an outcome, can
be held constant when the same household is compared over time.
Change in an outcome over time is measured by comparing the value
of the outcome indicator at two different points in time. In this
paper, change is defined as positive absolute difference or an
improvement in the outcome indicator. If the observed change (either
in absolute or in relative terms) in the outcome indicator for
participating households is greater than the observed change for
non-participating households, then the positive difference can
be attributed to the positive effect of program participation
on that outcome. If the magnitudes of observed change are not
different even then the absence of a difference can be attributed
to the equalising effect of participation, which implies that
participation has the effect of reducing socio-economic differences
in the outcome variable. Similarly, effects can be attributed
to the regularity of participation by comparing the difference
for continuous and occasional participants.
Change over time in social well-being and women’s position is
measured by actual differences in outcomes and/or processes between
the first and the third rounds for each individual household that
was available for interview in both the rounds. For comparing
change over time, relative difference, in terms of better, same
or worse, rather than absolute difference was considered to be
a more useful comparison, and indeed for the majority of the indicators,
only relative difference was computable. Since the assumption
is that participation will increase social well-being, change
will be described only in terms of improvement (better) and no
improvement (same or worse) in the selected indicators. Only in
the case of children’s school enrollment, improvement also includes
the category of same, which indicates that none of the children
dropped out from school, and
therefore, is a positive change.
The comparison of change
over time, i.e., improvement or no improvement, in the outcome
and process variables can be presented in more than one way. First,
the comparison can be between the mean value of the outcome in
the first and third rounds according to household participation
status. This gives the average change in the group over time,
but without accounting for the effect of non-program household
and village factors and unobserved selectivity factors on the
outcome or the process. The other comparison is the difference
or change in outcome for the same household, with the mean difference
in outcome being compared between the three participation groups.
In this case the difference in outcome, if any, can be attributed
to program participation, since the effect of all household and
village level factors, not affected by program participation,
are the same in both rounds. However, in this case the effect
of exogenous non-program factors, like policy change or change
in the macro-economic environment or natural disasters, that are
not present in both rounds, may distort the measurement of change
over time.
The comparison of change in
outcomes and processes will be between early 1998 (the round one
survey) and early 2000 (the third round survey) and will be restricted
to households that were available for interview in both those
rounds. The question of bias in mean estimates due to the possibility
that households are missing in the final survey in a systematic
fashion, a common problem of panel comparisons, cannot be ruled
out. However, of the 3026 households interviewed in the third
round, 2903 or 96 per cent were available in all three
rounds. Since the proportion of missing households was
actually quite small, this should not bias estimates a great deal
even if households were missed non-randomly. Since many households
have undergone division into extensions, aggregation was necessary
to make appropriate comparisons possible. This will also introduce
some bias since the aggregation of extension households is not,
strictly speaking, the same as the original undivided household.
Aggregation of outcome variables has been either as sum or mean
as relevant, and, where this is not possible (for example, in
the case of variables like contraceptive use or decision making),
only the parent household in the final round has been used for
comparison. In all cases comparisons were made on the basis of
averages, mean proportions and percentages.
6.3
Socio-economic Profile of Sample Households
It is not unlikely that
the socio-economic characteristics of sample households will vary
according to the nature of participation in a micro-credit program.
Almost by definition participant households are different from
non-participants, and continuous participant households are different
from occasional participants with respect to a large number of
observed indicators of socio-economic status. There are also significant
unobserved differences that influence selection of households
into or out of programs, and that influence the propensity of
participating households to drop out and rejoin programs. Socio-status
is described in terms of household demographic status, initial
size of land owned, initial occupation and education level of
head and initial household poverty status. The variations in membership
characteristics for participating households are described in
terms of sex of member, duration of membership, multiple membership
and current loan status. These are shown in Annex Tables 6.A.1-6.A.8.
The demographic status of households
shows considerable variation. Continuous participant households
have a larger family size compared to the other groups, and are
more likely to have both an adult male and an adult female family
member and to be liable to division. Household heads in continuous
and occasional participant households are more likely to be currently
married and male, and somewhat younger compared to heads in non-participant
households. Heads in all participant households are less likely
to be able to read and write and to have a lower level of schooling
compared to heads in non-participant households.
Not surprisingly, average size
of own land is the lowest for continuous participant households
and highest for non-participant households. Moreover, participant
households generally have a lower size of operated land and own
a smaller homestead than non-participant households. Heads in
continuous participant households are more likely to have a non-agricultural
occupation and less likely to have agricultural or service occupations
than occasional and non-participant households.
Initial poverty status is indicated
by the ‘eligibility’ criterion, household food availability, household’s
need for micro-credit, recent sale of land and the presence of
female wage labour. Continuous participant households are more
likely to conform to the NGO eligibility criterion compared to
the other groups, and occasional participant households are more
likely to fulfill the criteria than non-participant households.
A fairly large proportion of all households reported food deficit
at some time during the year, but this figure was higher among
participants compared to non-participants. Non-participants were
also more likely to be food surplus households compared to participants.
The need for credit was expressed by continuous participant household
more commonly than occasional participant households. Continuous
participants were more likely to rely on female wage labour than
the other household groups. Hence, the initial poverty and credit
need of continuous participant households was greater compared
to the other groups, and the initial poverty of occasional participants
was somewhat greater than non-participants.
The socio-economic backgrounds
show that with respect to household livelihood and poverty indicators
occasional participant households are more akin to non-participant
households than to continuous participants, which may be the reason
for their irregular participation. This is also reflected in the
membership characteristics, which reveal that continuous participant
households are more likely to have joined earlier and to have
more than one family member enrolled into a program. They are
also more likely to be members of more than one organization and
to have a current NGO loan.
6.4
Effect of Participation on Social Well-being
The
objective of this chapter is to assess the effect of participation
in micro-credit programs on household social well-being and vulnerability.
Social wellbeing will be indicated by social outcomes at the household
level, including household quality of life, children’s access
to schooling, family members’ access to health care and child
immunisation, expenditures on health care, schooling and house
repair, and vulnerability to crises and coping strategies adopted.
Respondents‘ perceptions of change in the indicators of social
well-being will also be examined. Wherever relevant, gender differences
will be reported to examine the possibility of gender inequalities
in well-being.
Quality of Life
One of the most desired effects of micro-credit program participation
on household social well-being are impacts on the quality of life.
The hypothesis suggests that since access to credit increases
household income program participation is likely to increase in
investment on household living standard leading to improvement
in the quality of life of household members. Household quality
of life is assessed by several living standard indicators, which
are the material of the main room, whether the household has electricity
connection, the type of toilet used by adult family members, and
the use of tubewell water for drinking and washing. The mean values
of these indicators in 1998 and 2000 are shown in annex Table
6.A.9a and 6.A.9b. The
proportion of sample households living in pucca huts is extremely
low (2-3%) generally and only occasional participant households
have a higher proportion of pucca huts compared to the other groups.
This proportion has not changed for any of the participation groups
between 1998 and 2000. The majority (over two thirds) of sample
households in all three groups live in tin huts and there has
been a visible increase in this proportion over time, which is
similar for all groups. The proportion with electricity connection
has also increased slightly between 1998 and 2000. In this case
some group difference is noted, with relatively greater increase
for non-participant households followed by continuous participants.
The use of sanitary toilet has become
much more common since 1998 for all sample households and for
women and men. This increase is similar by gender. There are differences
in the use of sanitary toilet by participation group, with occasional
participants being relatively more likely to use sanitary toilets
compared to the other household groups. On the other hand, the
use of open space by adults for toilet purpose has shown some
decline over time for all groups, the decline being largest for
occasional participants and continuous participants. No change
over time is noted in the extent of use of tubewell water for
drinking, which was fairly high among all groups, but there are
some group differences that have not diminished over time. In
this respect occasional participants have a lower than average
proportion of households using tubewell water for drinking. The
use of tubewell water for washing has increased slightly for all
groups and, as with drinking water, occasional participants fare
worse than the others. This is somewhat unexpected since these
households are relatively better-off in socio-economic status
compared to continuous participant households. One explanation
is that regular participation reduces the socio-economic difference
in water use behaviour.
The magnitude of change over time in the indicators of living
standard reflecting the quality of life is shown in Table 6.1.
There has been improvement in the material of the main room (from
tin hut to brick, or from bamboo/thatch to tin) among 10-12 per
cent of households, with relatively greater likelihood of improvement
among occasional participants. Improvement in toilet use (from
open space to kutcha toilet and from kutcha to sanitary toilet)
by adult family members was observed to a relatively greater extent
(25-29% of households), and improvement was relatively more likely
among participating households compared to non-participant households.
Very few households in all three groups experienced improvement
in drinking water use (from other sources to tubewell water) because
the vast majority of households in all groups were already drinking
tubewell water. However, about 9-12 per cent of households reported
improvement in the use of washing water, and the extent of improvement
was slightly higher among non-participant households. A similar
pattern was seen with respect to the extent of improvement in
household electricity connection.
Table 6.1
Effect of Participation on Social
Well-being
Percentage of Households with Improvement in Selected Indicators
of LivingStandard between 1998 and 2000 According to Nature
of Participation
|
Nature
of
Participation
|
|
Material
of main room
|
Toilet
|
Drinking
Water
|
Washing
Water
|
Electricity
Connection
|
|
Continuous
Participant
|
%HH
|
10
|
29
|
1
|
9
|
7
|
|
N
|
1003
|
1058
|
1066
|
1066
|
1061
|
|
SD
|
.30
|
.46
|
.12
|
.29
|
.26
|
|
Occasional
Participant
|
%HH
|
12
|
29
|
1
|
10
|
7
|
|
N
|
871
|
942
|
950
|
950
|
948
|
|
SD
|
.32
|
.45
|
.11
|
.30
|
.26
|
|
Never
Participant
|
%HH
|
10
|
25
|
1
|
12
|
9
|
|
N
|
778
|
879
|
885
|
885
|
884
|
|
SD
|
.30
|
.43
|
.10
|
.10
|
.28
|
Note: N-Number of Observation
In summary, quality
of life has improved for a small proportion of households in all
the groups but the extent of improvement was not substantial,
except in the case of toilet use by adult family members. In this
case a part of the improvement can be attributed to participation
in a micro-credit program, which is plausible since programs actively
promote the use of sanitary toilets and may even provide loans
to construct sanitary toilets. Although the independent effect
of participation in improving the quality of life is not very
visible, participation is able to reduce socio-economic differentials
in the quality of life.
Children’s Access to Schooling
An important indicator
of social well-being is the extent to which children in the household
have access to schooling. Since there is an income effect on access
to schooling it is expected that program participation, which
increases household income, will lead to greater investment on
children’s education and increase access to schooling. Alternately,
even in the absence of increase in household income program inputs
to promote children’s education can also increase access to subsidized
and low cost schooling (non-formal education, food for education,
secondary school stipend program for girls, etc.). Moreover, it
is expected that the gender bias in access to schooling should
be reduced by program participation.
Table 6.2 presents
the mean proportion of school-age children currently enrolled
in school in 1998 and improvement in children’s access to school
between 1998 and 2000. Access to schooling is considered to improve
over time if there is an increase in the proportion of school-age
children enrolled or if there is no decline in this proportion,
i.e. there are no children dropping out of school. In the case
of children aged 6-9 years nearly two thirds were in school in
1998 in all household groups, with a reversed gender bias (i.e.
in favour of girls) in participating households compared to non-participating
households. The mean proportion of children aged 6-9 years currently
in school has increased over time, to the same extent for boys
and girls in most of the households in all groups, with a slightly
greater likelihood of increase among occasional participants.
However, unlike participating households the extent of increase
in the proportion of boys enrolled has been greater than the extent
of increase in the proportion of girls enrolled among non-participant
households. In other words, access to schooling for children 6-9
years old has increased in general and a participation effect
in increasing access to schooling is observed to the extent that
access depends upon household socio-economic status, which is
relatively lower for continuous participating households. However,
there is a clear participation effect in terms of the removal
of the gender gap in access to schooling, which actually persists
over time among non-participant households.
In the case of children aged 10-14 years old, the proportion currently
in school in 1998 was even higher (87-90 %of girls and 84-87 %
of boys) in all the household groups. As before, there was a reversed
gender bias in access to schooling in participating households,
which was not very visible among non-participating households.
For this age-group of children, however, although the gender bias
in access to schooling is not reversed it has been removed among
non-participating households. Almost all households in all the
groups experience increase in the proportion currently enrolled.
Hence, access to schooling for children 10-14 years old has also
generally increased over time and the participation effect on
access to schooling is seen in terms of equalizing the socio-economic
difference in access and reversing the gender bias.
Table 6.2
Proportion
of School-age Children Currently Enrolled and Percentage of Households
Experiencing no Decline/increase between 1998 and 2000 in the Proportion
of School-age
Children Enrolled
According to Nature of Participation
|
Nature
of
Participation
|
|
Children
6-9 years
|
Children 10-14 years
|
|
|
Proportion
of children enrolled in 1998 (Mean)
|
No
decline or increase in proportion enrolled (%HH)
|
Proportion
of children enrolled in 1998 (Mean)
|
No
decline or increase in proportion enrolled (%HH)
|
|
|
|
Girls
|
Boys
|
Girls
|
Boys
|
Girls
|
Boys
|
Girls
|
Boys
|
|
Continuous
Participant
|
|
.69
|
.66
|
88
|
88
|
.87
|
.84
|
97
|
99
|
|
N
|
327
|
350
|
198
|
217
|
378
|
411
|
288
|
334
|
|
SD
|
.44
|
.46
|
.33
|
.32
|
.32
|
.35
|
.18
|
.11
|
|
Occasional
Participant
|
|
.68
|
.63
|
90
|
90
|
.90
|
.85
|
98
|
96
|
|
N
|
255
|
287
|
157
|
174
|
331
|
320
|
247
|
255
|
|
SD
|
.46
|
.47
|
.29
|
.31
|
.30
|
.34
|
.15
|
.19
|
|
Never
Participant
|
|
.69
|
.70
|
87
|
92
|
.88
|
.87
|
97
|
99
|
|
N
|
231
|
265
|
138
|
167
|
272
|
279
|
215
|
204
|
|
SD
|
.45
|
.44
|
.34
|
.29
|
.31
|
.33
|
.17
|
.12
|
| |
