Causes of Fertility Decline in India and Bangladesh: Role of Community

Author(s): S. Mahendra Dev, K. S. James and Binayak Sen

Source: Economic and Political Weekly , Oct. 26 - Nov. 1, 2002, Vol. 37, No. 43 (Oct. 26 -
Nov. 1, 2002), pp. 4447-4454
Published by: Economic and Political Weekly

Stable URL: https://www.jstor.org/stable/4412781

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Economic and Political Weekly is collaborating with JSTOR to digitize, preserve and extend
access to Economic and Political Weekly

This content downloaded from

14.139.114.34 on Fri, 19 Mar 2021 06:47:43 UTC
All use subject to https://about.jstor.org/terms
 Causes of Fertility Decline in India
and Bangladesh
Role of Community
Conventional approaches to studies on fertility decline have long assumed the primacy of the
household as the prime decision-maker. Aspects of the household such as its .economic
standard of living, social standing, exposure to mass media, work status were some of the
influential factors at work on a couple's decision on their number of children. However,
individual and household level factors have been unable in several instances to explain the
full course of fertility transition seen in some Indian states and Bangladesh, where in some
regions fertility transition cuts across socio-economic and cultural boundaries.Gaps
in understanding such trends have been as this paper suggests due to the conventional
emphasis on household level variables. It argues instead for the need to look at the influence
the community plays in south Asia and to understand the levels of interaction that
exist at household level and at the community level.

S MAHENDRA DEV, K S JAMES, BINAYAK SEN

work status, etc, are thought to be some household level and at the community level.
Introduction of the influential factors driving couple's Our argument comes closer to the ques-
decision on number of children. tion addressed in sociological literature on
ne of the important social changes However, recent evidences suggest that micro-macro linkages between individual
taking place in developing coun-the individual and household level factors and community [Alexander 1988, Coleman
tries recently is the rapid decline are unable to explain the full course of 1990]. This question was also partially
in fertility defying conventional theoriesfertility transition taking place in some ofaddressed in the context of demographic
the states in India and Bangladesh. Studiesliterature [Bongaarts and Watkins 1996,
and beliefs. The fertility decline observed
in Bangladesh and many states in India,observed a geographical regularity in Entwisle et al 1989, Munshi and Myaux
thus, still remains a paradox. The decline,Indian fertility transition [Guilmoto 2000]. 2000]. The concept of diffusion of family
The fertility transition in some regions cutplanning was also based on the fact that
in this region, had taken place in extremely
difficult conditions, like, low standard ofacross socio-economic and cultural bound- there is a peculiar geographical pattern in
living, low literacy level, lack of moderni-aries. For instance, contraceptive use, infertility irrespective of socio-economic
sation, etc. states like Andhra Pradesh and even in background. However, we depart from the
The conventional approach to the stud- Bangladesh, was predominantly by those diffusion argument in the following way.
ies on fertility decline had been based on with low economic and social background.First, diffusion as a theory only speaks
the strong premise that the decision on the Although at a superficial level there may
about the spread of birth control technol-
number of children is taken purely pri- ogy. We argue that it is not the mere
be a tendency to call it as poverty induced
vately within the household decision- one, one need to go a little further information
to on the birth control, (which
making framework. Hence the question understand the real factors behind such
most of the women in India and Bangladesh
has been on how the individual decision important social transformation. possess even earlier) but is the sum total
on number of children is shaped in differ-We, therefore, argue that the failure of
of one's interaction within the community
ent settings? Both economic and social the conventional approach in explaining that makes one to take a progressive
development arguments emphasised their the recent decline in fertility is due to
decisions like the acceptance of contracep-
overemphasis on household level variables.
respective roles in shaping such individual tive use. Secondly, the diffusion does not
Thus, though purely private, decision on
decisions. In other words, the attempt thus speak about the pathways through which
far was to assess the importance of indi-the number of children and contraceptivethe acquired knowledge on birth control
vidual or household level variables in use are mostly shaped at the level of
is put into practice. We argue that the
community in the context of south Asian
influencing fertility decisions. This assess- diffusion takes place through a social
interactive framework (such as associa-
region. Hence, the impelling force behind
ment is partly due to the fact that fertility
theories are drawn mainly from the a progressive decision like contraceptive
expe- tional activities, self-help groups, infor-
rience of western European countries. acceptance
Thus often is the level of one'smal gatherings around, etc) rather than
household economic standard of living,interaction
its within the community. Thus through a purely individualised interac-
social standing, exposure to mass media,
we try to integrate the argument at the
tion. Third, while claims are being made

Economic and Political Weekly October 26, 2002 4447

This content downloaded from

14.139.114.34 on Fri, 19 Mar 2021 06:47:43 UTC
All use subject to https://about.jstor.org/terms
in the literature about the geographical fertility decline observed in Kerala region
and had been emphasised in such cases.
diffusion by method of exclusion, the other few developing countries likeASri pioneering explanation lies in this con-
positive test for existence of such geo- Lanka, Costa Rica, etc. All these places
text was the supply side argument consid-
graphic effects was not carried out. In this ering that vigorous implementation of
had a decline in fertility unaccompanied
study, we attempt to capture the geographic by adequate economic development, family planning itself could generate de-
effects through an improved statistical industrialisation orurbanisation. However,
mand for such services [Anthony 1992,
analysis. However, it may be mentioned invariably two striking observationsSrinivasan
are 1995]. Such explanations were
that while the statistical models point to strongly brought out in the context of
evident from the experience of these coun-
the definitive existence of such effects, it Bangladesh and Tamil Nadu. However, it
tries. First, all these countries had very
falls short of pinpointing the policies and/ is doubtful as to whether supply side
high overall literacy rate and particularly
or initial conditions that contributes for the high female literacy. Second, the health
argument can provide a general frame-
making of such geographical effects. status of the people mainly the infant work to explain contemporary fertility
The objective of this paper, therefore, transition. There is ample empirical evi-
mortality rate was strikingly lower than
is to bring out community as context for any other places [Krishnan 1976, Bhatdence
and to the fact that the supply of con-
fertility transition like India and Bangladesh Rajan 1990]. Hence attention was shifted
traceptive without generating adequate
without delimiting the importance of from economic argument to a social de- demand often remained unsuccessful.
conventional socio-economic factors. To velopment discourse including health In this context, the diffusion hypothesis,
model these effects data are drawn from improvement. This argument had been offered as an explanation of fertility tran-
Demographic Health Surveys (known as mainly accepted until the 1980s. sition also needs particular mention. Dif-
National Family Health Survey in India). It was then followed by a sudden decline fusion mainly speaks about the spread of
As conventional statistical techniques fall in fertility in Tamil Nadu and later in control technology.
birth
short of recognising the multilevel struc- Andhra Pradesh without improvement A insimilar explanation has been offered
ture under examination, household and female literacy or decline in IMR. Thisfrom has the experience of several developing
community, we adopt an improved multi- led to the understanding that there is countries;
no 'social interaction' has been
level modelling especially to capture the threshold level of female education or infant
suggested as one of the powerful forces
community effect in fertility transition. mortality that can be considered for aug-
that accelerate the pace of fertility change
inisa society. [Bongaarts and Watkins 1996].
mentation of fertility decline. The same
II There can be different channels of social
also proved from the experience of other
states in India such as Karnataka and West interaction such as spatial proximity to
Background
Bengal and also from Bangladesh. urban centres, extensive media facility
With the demographic transition theory However, the emphasis has been to bring and so on.
thought to be intact in explaining fertility out the importance of other household An alternative to such explanation but
transition in any society, developed and level variables other than female literacy confining to macro changes was based on
developing, rapid pace of fertility decline and infant mortality. The onset of the the conducive atmosphere of a particular
is expected only with a significant eco- electronic era has also persuaded the re- geographical or rather administrative re-
nomic and social change. As a result, there searchers to emphasise the importance of gion created mainly by governmental o
will be rising costs and declining eco- the media. It is argued that while in Kerala other efforts. For instance, in the case of
nomic values of children driving couples the print media played a crucial role, in Kerala social justice in political and eco-
to go for smaller families. The other major Tamil Nadu, and to some extent in Andhra nomic policies and development strategie
economic argument at the micro level also Pradesh, the access to electronic media reflected through land reforms and in-
emphasised the conflict between quality was crucial in explaining fertility decline creased wages to agricultural labourer
and quantity with rising income.They also [Bhat 1996]. [Ratcliffe 1978], success of backward clas
anticipated rising opportunities for women As against the household model there movements in the state [Anandhi 1996,
with 'time' becoming an important factor had been an attempt to project macro Srinivasan et al 1991], populist policies
in household decision on number of chil- changes leading to larger fertility transi- of the government in Andhra Pradesh
dren. However, as far as developing coun- tion. The crux of such argument is that [James 1999]. A major drawback of this
tries are concerned such modernisation or policies and programmes undertaken or argument is that fertility transition is viewed
economic factors do not seem to explainthe changes occurring in a particular geo- as bits and pieces rather than in a genera
the full course of fertility transition. First,graphical area would possibly have the framework. It then shows that the transi-
the fertility decline has occurred in manyability to transform the entire geographical tion in each geographical region needs to
countries and areas where the levels of area including the fertility decisions of the be explained uniquely. Further with fer-
economic development were comparatively couples. Hence, such change embraces all tility data available from India on smaller
poor. Second, the opportunities to ensure those living in a defined geographical area geographical units than state (like district)
the quality of children are limited inirrespective
the of the socio-economic char- it became increasingly clear that such ex-
context of many developing countries. acteristics. The pattern of fertility transi- planations do not hold good any more.
tion, therefore, is horizontal in nature than
Third, the value of time is not particularly The other attempt has been to place the
relevant when majority of women vertical
are in the sense that the place of fertility transition in the larger context of
illiterate and underemployed. residence that matters than the socio-eco- history, culture, or the policies and
nomic
The empirical validity of the 'role of standing. Thus, rather than the programmes of the government. In the
individual or household characteristics the
economic factors in augmenting fertility government programmes there were two
transition had been questioned with people
the living in a particular geographical major arguments. First was the efficient

4448 Economic and Political Weekly October 26, 2002

This content downloaded from

14.139.114.34 on Fri, 19 Mar 2021 06:47:43 UTC
All use subject to https://about.jstor.org/terms
supply of contraceptives by the govern- north coastal region has also recorded the It does not mean that fertility variation
ment even when there is lack of sufficient lowest fertility rate. The argument on the is not there between different socio-
demand generated in the society. This will geographical contiluity in fertility decline economic groups in fast fertility declining
trigger off an increase in the acceptancealso primarily emanates from the re- regions. Although the magnitude of dif-
rate. The second was the policies of the gional or district level data available froml ference is comparatively less, fertility
government in the social and economic different states in India. This firmly un- differences persists in all population ir-
sector that trigger off social change includ- derplays the importance of socio-economic respective of the pace of decline in overall
ing the fertility transition. These arguments, indicators to trigger fertility transition. fertility. Table 3 presents total fertility rate
therefore, consider variables not at the This is not only true with regard to India by background characteristics of ever-
household level but at a macro or com- but also in the neighbouring country married women in India, Bangladesh and
munity level. However, it is rather easy for Bangladesh. It is observed from district Andhra Pradesh.
level studies that contraceptive use is
these studies to find out some regularity The table clearly brings out the fertility
between fertility decline and governmen-
significantly high among those districts differentials existing by level of education,
tal actions but could not conclusively bordering the state of West Bengal in religion, caste, standard of living index,
establish any causal link mainly due which to the same language is spoken [Aminetc. Such differences persist in all the places,
inadequacy of statistical techniques. et al 2000]. In other words, rather thanthough, amazingly the magnitude of such
However, many of the conventional socio-economic continuity, it was a sort variation is considerably different. While
factors which tried to explain fertility of geographical contiguity that was ob- the total fertility rate of illiterates in India
transition in the past become insignificant served in the case of West Bengal and and Bangladesh is nearly 3.5 and 3.9
Bangladesh which was explained in terms children respectively that in Andhra
toexplain dramatic decline in fertility taking
place in many developing countries today. of its language, culture and history [Basu Pradesh is only 2.4. It may be recalled that
In this paper, we search for explanations and Amin 2000]. nearly 64 per cent of ever-married women
beyond individual or family characteris-
tics in explaining fertility decline. Table 1: TFR, Female Literacy and IMR for Various States in India

III States TFR 1996-98 Female Literacy Rate, 2001 Infant Mortality Rate1998-99
Andhra Pradesh
Fertility Pattern Assam 2.3 56.03 69.5
Bihar 3.5 33.57 72.9
Fertility decline, in fact, can be observed
Goa 1.8 75.51 36.7
in many states in India, though the accel- Gujarat 2.7 58.60 62.6
erated pace of transition is restricted to a Haryana 2.9 56.31 56.8
Himachal Pradesh 2.1 68.08 34.4
few states. Thus the country has now
Karnataka 2.1 57.45 51.5
considerable regional heterogeneity in Kerala 2.0 87.86 16.3
demographic characteristics. Table 1 pre- Madhya Pradesh 3.3 50.28 86.1
sents total fertility rate in different states Maharastra 2.5 67.51 43.7
Orissa 2.5 50.97 81.0
in India during 1996-98 period. Two other
Punjab 2.2 63.55 57.1
conventional indicators that are thought to Rajastan 3.9 44.34 80.4
be crucial in fertility transition are also Tamil Nadu 2.2 64.55 48.2
presented in the table. Uttar Pradesh 4.0 42.98 86.7
Over two child differences exist between West Bengal 2.3 60.22 48.7
2.3 51.17 65.8 India
low and high fertility states in India. The 2.9 54.16 67.6
association between conventional factors
Note: IMR is for five year
and fertility is also fairly strong consi-
Source: TFR and IMR data a
dering a state level analysis. The correla- census of India (20001)
tion coefficient of Total Fertility Rate (TFR)
with Infant Mortality Rate (IMR) as well Table 2: TFR, Fema
of Andhra Pradesh
as female literacy is over 0.7 indicating
strong relationship. However, such rela- Region TFR (0-9 Years Percentage of Literate Child (under Age 5)
tionship does not hold good if one look Preceding the Survey, Ever-Married Women Mortality, Approximately
1992-93) (13-49 years) for 1986-88 Period
at regional or sub-regional fertility levels
and socio-economic variables in rapidly North Coast 2.9 23.1 148
fertility declining states. For instance in South Coast 3.0 41.0 117
Rayalaseema 3.7 22.7 132
Andhra Pradesh those districts with poor Telangana 3.4 31.0 96
literacy rate in the coastal area like Godavari depression 3.0 26.5 114
Srikakulam and Vizianagaram also re- Andhra Pradesh 3.2 31.3 117
corded steep decline in fertility. The re- Note: The following are the districts
gional level data for Andhra Pradesh is North coast: Srikakulam, Vizianag
presented in Table 2. South coast: East Godavari, West G
Rayalaseema: Chittoor, Cuddapah,
It is a bit puzzling that even with the
Telangana: Mahabubnagar, Rangare
percentage of literate as low as 23 per cent Godavari Depression: Karimnagar,
and with highest child mortality rate the Source: Bhat and Xavier (1999).

Economic and Political Weekly October 26, 2002 4449

This content downloaded from

14.139.114.34 on Fri, 19 Mar 2021 06:47:43 UTC
All use subject to https://about.jstor.org/terms
in Andhra Pradesh is illiterate according bi-variate regression is summarised in explanation until today on the geographi-
to the National Family Health Survey, Table 5. cal analysis of fertility was the diffusion
1998-99. The fertility differentials are sub- It is clear from the table that many theory.
of
stantially smaller by different background the socio-economic variables important in
characteristics in the state. Surprisingly, explaining variation in fertility at commu- V
both households with low and medium nity/ district level earlier turned out to be
Concept of 'Diffusion'
standard of living have the same total insignificant in many states and regions for
fertility rate. However, in India the and1991 census year. For instance, female The essence of diffusion argument is
Bangladesh the differences between dif-literacy was not found important in states that birth control is seen as an innovation
like Kerala, Tamil Nadu and Andhra
ferent educational groups in fertility rate that spreads from person to person, from
are sizeable. Pradesh. Bangladesh in this respect group to group and region to region [Van
shows similar pattern as West Bengal. It De Kaa 19961. Obviously, the major evi-
IV is understandable in the case of Kerala as dence on diffusion is garnered from west-
Conventional Factors female literacy is nearly universal and
ern Europe. The Princeton European Fer-
there is no significant difference acrosstility Project and other similar historical
Influencing Fertility Decline
districts. The variations in female literacy studies have established that ftrtility tran-
Table 4 presents some of the socio-in Andhra Pradesh and Tamil Nadu are, sition occurred within rather than across
economic and health indicators and totalin fact, striking. linguistic boundaries [Coale and Watkins
19961. However, the recent emergence of
fertility rate in India and Bangladesh andIn a nutshell, though relevant, the con-
some of the Indian states where there had ventional socio-economic factors become interest in diffusion is primarily the result
been a rapid decline in fertility in the 1980spoor predictors of contemporary fertility
of decline in fertility in many less devel-
and 1990s. oped countries within a brief time span.
transition. Rather the evidence points to
Some important conclusions canthe This
beimportance of particular geographical decline occurred with little or no
change in the conventional economic and
area in fertility transition. For instance, in
directly drawn from this table. First, not-
withstanding the significant variation incase of fast fertility declining states likesocial indicators.This unprecedented de-
the
socio-economic circumstances there has Andhra Pradesh, none of the conventional cline in fertility in less developed countries
been a significant reduction in fertility in socio-economic factors seem particularlyunrelated to economic and social factors
these states. Bangladesh's poverty level is relevant for explaining district level fer- led Retherford (1985) and later Cleland
twice as high as India, but the observed tility variation. However, even with con- and Wilson (1987) to emphasise the role
level of fertility for the latest available siderable adversities the north coastal of diffusion process of family planning.
years displays an identical scenario. The districts in the state underwent rapid pace They cannot, however, distinguish the
same applies when fertility levels are of fertility decline. This leads us to effect
un- of spread of knowledge and ideas
compared among the rapidly fertility- derstand more on the geographicalfrom di- the structural determinism [Van De
Kaa 1996].
declining states. The head-count index of mension of fertility transition. The major
poverty varies from 15 per cent in Andhra
Pradesh to 27 percent in West Bengal with Table 3: TFR by Different Background Characteristics in India
nearly identical TFR. In short, there has and Bangladesh
been a very substantial decline in fertility
Characteristics Total Fertility Rate
in the context of the low level of per capita India Andhra Pradesh Bangladesh
income and high level of income-poverty. 1996-98 1996-98 1996-97

This is also observed with regard to Residence

urbanisation or percentage of workers Urban 2.27 2.07 2.10

outside agriculture. Female literacy seem Rural 3.07 2.32 3.43

Education
to have a major role in Kerala and to some Illiterate 3.47 2.35 3.93
extent in Tamil Nadu but not in other states
Literate, < middle complete 2.64 2.22 3.27
like Andhra Pradesh, Karnataka or Middle school complete 2.26 1.94 3.01
High school and above 1.99 2.20 2.12
Bangladesh where the rates are still quite
Religion
low. Even infant mortality rates are quite
Hindu 2.78 2.20
high in many states,judged by the standard
Muslims 3.59 2.53
Christian
of Kerala, where the fertility transition is 2.44 2.53
observed in the last two decades. Sikh 2.26
Caste/tribe
The relationship between fertility and its
Scheduled caste 3.15 2.51
determinants has been examined using Scheduled tribe 3.06 2.75
Other Backward Classes 2.83 ' 2.26
regression analysis for the above states
Other 2.66 2.00
using district level data from the 1991
Standard of living index
Census [Census of India 1991]. For Bangla-Low 3.37 2.29
desh we have used the district-level data- Medium 2.85 2.32
base developed for the Bangladesh Hu- High 2.10 1.99
Total 2.85 2.25
man Development Report 2000 carried
out by Bangladesh Institute of Develop- Source: For India and A
ment Studies (BIDS). The results of the for Bangladesh from

4450 Economic and Political Weekly October 26, 2002

This content downloaded from

14.139.114.34 on Fri, 19 Mar 2021 06:47:43 UTC
All use subject to https://about.jstor.org/terms
 There are several types of diffusion: technology and why some persistently The influence of community on indi-
social, geographic or both [Bocquet-Appel ignore neighbour's experience [Munshi vidual behaviour suggests that the com-
and Jakobi 1998]. Social diffusion occurs and Myaux 2000]. Bongaarts and Watkins munity often has a major role to play in
through different means, education, mass (1996) concluded that social interaction is private decisions. Due to various
certain
media, government efforts and so on. It is constraints and societal attitude woman as
a critical process in fertility transition.
yet to establish clearly which .of these Based on the study conducted in Kenya an individual would be unable to take a
programmes pay the most dividends in they argued that much of social interaction
decision on acceptance of modern contra-
terms of fertility reduction. However, the on a day-to-day basis occurs in the context
ception. On thecontrary, becoming a part
essence of diffusion argument primarily of personal network, the building blocks
of larger groups provides them an oppor-
lies in geographical diffusion process than of communities. tunity to discuss and learn from other's
the social diffusion. experience and take progressive decisions.
The interest, in recent years, has been VI Bongaarts and Watkins (1996) observed
mostly to understand the geographical that social influence was a critical factor
Rethinking Diffusion
diffusion process. Geographical diffusion, in maintaining high fertility in may soci-
in a way, reveals the possibility of fertility In light of the above discussion there is eties. Social interaction gives a social
decline even among underprivileged (botha need for rethinking on the diffusion approval for certain individual behaviour.
economically and socially) if a womanmodel as currently discussed in the litera- The groups, however, are not formed
happened to live in a particular geographi-
ture. We argue that the defining diffusion with an objective to promote contraceptive
cal area. There are two pathways by whichpurely in terms of spread of birth control use. However, many of such programmes
geographic diffusion embraces a particu-technology from centre to periphery ig- target women. The experience of India and
lar area - (i) Social learning (ii) Social
nores the wider context of social change Bangladesh suggest that the main channels
Influence (Montgomery and Casterline occurring in south Asian Societies. Thus of social interaction are self-help groups.
1998). Social influence refers to the power
we define diffusion process in a larger These groups were created with a com-
that individuals exercise over each other context of individual-community inter- pletely different motive. However, the
through authority, deference and social action, closer to one put forward by existence of such groups is also expected
conformity pressure [Montgomery and Bongaarts and Watkins (1996). Though to have a major impact on those decisions
Casterline 1998]. Although the possibility
purely private, decision on the number of like contraceptive acceptance. In India, for
of government authority in the implemen- children and contraceptive use are often instance, efficient functioning of self-help
tation of family planning is often talkedshaped by one's interaction within the groups can be observed in those states which
about in Indian context, it is now largely
community. is undergoing rapid fertility transition.
regarded that cohesion does, not play a
major role in family decisions. Social Table 5: Results of Bi-variate Regression of TFR on Selected
Socio-Economic Variables for 1991
learning essentially describes a process by
which the individual learns from his
Country/State Female Female Female Working Percentage Percentage Infant
neighbours' technology. That is, the infor- Literacy Working in Secondary of Pucca of Urban Mortality
mation or behaviour of one person can Participation and Tertiary House Population Rate
Sector
have spillover effects on the motivation of
another. This can have either positiveBangladesh
or - - *
India
negative impact [Montgomery and
Andhra Pradesh - -
Casterline 1998]. Karnataka - -

It is not yet clearly understood the dif-

Kerala
Tamil Nadu - - * *
ferent pathways by which diffusion pro-
West Bengal * - * *
cess operates in a particular geographical
area. Diffusion also does not speak about
Note: * Significant at 5 per cent leve
the pace of adoption of birth control Source: For India and states from Ce
developed for the Bangladesh Hum

Table 4: Conventional Factors Infl

Country/States Headcount index Percentage of Percentage of Percentage of Percentage of Percentage of IMR TFR
(Per Cent) Urban Female Female Women not Workers
Population Literates Workers Exposed to any Outside
Mass Media Agriculture
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Bangladesh 46.0 20.0 41.50 15.0 54.3 63.1 65 3.3
India 26.1 25.7 54.16 22.3 40.3 32.5 71 3.3
Andhra Pradesh 15.8 26.9 51.17 34.3 23.7 28.8 63 2.5
Karnataka 20.0 30.9 57.45 29.4 21.4 32.6 53 2.5
Kerala 12.7 26.4 87.86 15.8 11.5 52.0 12 1.8
Tamil Nadu 21.1 34.2 64.55 29.9 20.3 38.2 53 2.0
West Bengal 27.0 27.5 60.22 11.2 38.6 43.5 55 2.6
Note: Latest available data.

Source: For India column no 2 is from Planning Commission estimates for 1999-2000, 3, 5 and 7 is from Census of India (1991), 4 from Cen
6, 8 and 9 is from National Family Health Survey 1998-99. For Bangladesh these figures are compiled as part of the Bangladesh Hu
Report (2000).

Economic and Political Weekly October 26, 2002 4451

This content downloaded from

14.139.114.34 on Fri, 19 Mar 2021 06:47:43 UTC
All use subject to https://about.jstor.org/terms
 The self-help group, DWCRA, for in- fect? Second, if such geographical effectthree level structure women at the level-
stance, is an undoubted success story in persists, there is a need to test the impor- I nested within primary sampling unit
Andhra Pradesh and formed a central tance of community level interaction as (PSU)a at the level-2 and PSU nested within
element in the state's community the district at level-3 is considered for
possible explanation for fertility transi-
mobilisation. Nearly 400,000 DWCRA
tion. Hence, the determinants to fertilityBangladesh. District level is omitted in
are hypothesised to operate at two levels:India as it is not considered in the sampling
groups are formed in the state constituting
individual-level (such as women's socio-frame of Indian National Family Health
nearly 40 per cent of total groups in all
economic status) and village-level (theSurvey.
India. Most of these women are of younger
ages with average age around 35 and alsogeographical/community factor). For in- The number of children born to each of
belong to socio-economically weaker sec- stance, a village could have high (low) the ever-married women is considered as
tions of the society. Not only this, such levels of fertility simply because it has adependent variables. Most of the conven-
groups helped women to earn additional concentration of certain types of womentional socio-economic variables are also
income but also would have helped them who anyway are likely to report high (low)considered here. Due to high correspon-
in providing a progressive outlook to such levels. In such cases, the variation between dence between female and male education,
issues on small family size and contracep- villages would simply be an artefact of its only female educational level is consi-
tive use. The evaluation survey conducted population composition. However, if dif- dered. For female education the reference
in Andhra Pradesh showed that one of the ferences between villages remain, even category is taken as high school and above
main issues discussed in the groups is after considering all possible individual for India and secondary and above for
health/sanitation and family planning [Org- factors that influence fertility (poverty, Bangladesh. Hence, the coefficient for
Marg Research Ltd 2000]. illiteracy, religion, caste and so on) then illiterate or primary level education indi-
In the case of Bangladesh, the Grameen it suggests that over and above the indi- cates the differences between them and a
Bank of Bangladesh is perhaps the best- vidual factors, there are 'village-effects'. high school level and above educated
known example of promoting self-help Consequently, it assesses whether or not women. Also religion, caste, media expo-
groups and there are millions of benefi- living in one village as compared to sure work participation, etc, are also con-
ciaries of which nearly 95 per cent are another has an impact on women's fertility sidered for the analysis.
women. There are also other credit ad- decision.' Conventional techniques fall We have also tested for a few village
vancing organisations like Bangladesh short of recognising the 'multilevel' struc- level variables for India representing group
Rural Advancement Committee (BRAC), ture of the issue under examination. interaction such as existence of women's
Bangladesh Rural Development Boards However, in recent times multilevel modell- associations in the villages or the number
(BRDB), etc. These programmes demon- ing procedures have allowed us to address of group meetings of family planning held
these concerns in a robust manner
strated that it is possible for group based in villages to test our basic hypothesis on
lending programmes to generate sustain- [Goldstein 1995]. the importance of community impact on
able livelihoods for the poor. These groups the fertility decision. A variable that may
meet weekly repayment of their loans thus Data and Methodology forroughly the indicate the conventional diffu-
enhancing the interaction within the com- Multilevel Model sion argument, distance to the nearest town
munity. Further, these programmes encour- from the village is also included as a village
age their members to have small families level predictor. This is based on the as-
This paper adopts the multilevel tech-
and educate their children. sumption
nique to quantitatively test the presence of that closeness to the town pro-
vides easy passage to the knowledge on
geographic effects. The data for the study
VII are drawn from the Demographic Health
birth control technology. As village mod-
Measurement Issues Surveys (DHS) conducted in India ule
andis not available for Bangladesh no
Bangladesh. In India the survey is known
variable is tested at the village and district
Although the existence of community as National Family Health Survey. As level.
the
level process in fertility decline is now As already pointed out, the important
latest round data (1998-99) is not yet made
largely accepted, the major questionavailable,
is we used 1992-93 data for India
question addressed here is how far various
how this can be effectively measuredand or correspondingly 1993-94 data for household level variables are capable of
Bangladesh.
understood. By plotting fertility variation explaining fertility variation in India and
in different areas, it is possible to assessIndian NFHS data contain different
Bangladesh. After duly accounting for the
the existence of geographical patternmodules,
in household level variables, whether there
household, women and village.
We have combined both women and vil-
fertility. However, such attempts provide is any significant variation remains at the
lage module for the multilevel analysis.
only an indication of geographical effect. village or district level.
The attempt in India using data at the The variables on village community
interaction are taken from' the village Discussion of the Results
district and village level has clearly re-
module. As a result, for India data consists
vealed spatial contiguity in the fertility
transition [Guilmoto 2000, Guilmoto and only for rural areas. However for Table 6 presents results of a multilevel
Rajan 2001]. Bangladesh only women's module is used model for India and Bangladesh. Most of
There are two measurement issues this for the analysis. the socio-economic factors at the house-
paper confronts. First it needs to assess For the multilevel analysis, a two-level hold level are significantly associated with
after controlling for conventional socio-structure of women at the level-1 nested the number of children both in India and
economic factors, whether there is any within the villages at the level-2 is consider-Bangladesh. The contraceptive use both in
significant geographical/community ef- ed for India. For Bangladesh, however, India and Bangladesh turned out to be

4452 Economic and Political Weekly October 26, 2002

This content downloaded from

14.139.114.34 on Fri, 19 Mar 2021 06:47:43 UTC
All use subject to https://about.jstor.org/terms
positively associated with number of were largely confined to either regional contraceptive use in that community irre-
children born. This was probably because specificities or the often-debated diffusion spective of the development in that par-
women in these countries use contracep- theory arguing the spread of birth control ticular geographical area. We also tried
tion mainly as a permanent method after technology. However, both these argu- to measure this relationship through an
achieving their desired family level. Illit- ments rather than finding out some regu- improved multilevel modelling.
erate and primary educated women have larity could not conclusively establish a The results showed, although substan-
nearly half the children higher than high causal link with fertility transition. This tial variation in fertility remains at the
school and above educated women in India. was also partly due to non-availability of household or women level, there is signifi-
Nearly similar picture is found in Andhra proper statistical tools to establish such cant unexplained variation at the village/
Pradesh and Bangladesh. link. district level. This confirms the fact that
As these results are fairy well known, This paper, however, argues that in the conventional theories are incapable of
where the present analysis makes a depar- context of developing countries like India explaining the recent rapid course of ferti-
ture and improves is by estimating the and Bangladesh the decision on contracep- lity transition in many developing regions.
'district effect' and 'village effect' on tive acceptance are often shaped at the We have also tested for the impact of
women's fertility. Random part of the community level. Being a part of commu- community interaction variable on the
model suggests that individual/household nity woman as an individual looks at the fertility decisions and found it significant.
level factors do not account for the total community response while taking a pro- This gives credence to our argument that
variation in the fertility. There is a signifi-gressive decision like contraceptive ac- the pace of fertility transition could be
cant variation between districts in Bangla-ceptance. If the social response is augmented through proper channels of
desh and between primary sampling unitfavourable a woman would accept family community interaction within a geo-
(PSU) variations in both the countries. Inplanning irrespective of socio-economic graphical area.
other words, poor, illiterate, scheduledstatus. We also argued that the individual However, it should be mentioned here
caste, etc, living in a particular village ordecisions are formed through a social that the village level factors considered
district seems to have significant impactinteractive process within the community. here are only instruments in explaining
on the fertility performance of the women.Hence better the opportunity to have com- community interaction. More needs to be
This evidence indirectly suggests themunity interaction, higher will be the understood as how some decisions are
importance of geographical region. This
evidence is crucial as it makes a case to
Table 6: The Multilevel Regression Coefficient and Standard Error of Eac
develop contextual explanations to explain Independent Variable on Children Ever Born in India and Bangladesh
variation in fertility levels
Dependent=Children Born India (1992/93) Bangladesh (1993/94)
As already pointed out some of the
Variables No. of cases=52843 No. of cases=9634
contextual variables are considered for the
Indian data to test whether the existence Women level predictors
Age of female in years 0.367 (0.005)* 0.228 (0.011)*
of a women's group or group level activi-Age Square -0.004 (0.000)* -0.001 (0.000)*
ties conducted in the village last one yearMale children died 0.937 (0.011)* 0.921 (0.023)*
have any impact on the fertility perfor-Female children died 1.008 (0.011)* 0.973 (0.023)*
mance of the women in that village. These Modern Contraceptive Use (1=yes; 0=No) 0.562 (0.014)* 0.681 (0.031)*
Women illiterate (reference category=
two variables also show a significant re-
High school & above for India and
lationship with fertility. However, theSecondary and above for Bangladesh) 0.733 (0.030)* 0.453 (0.046)*
variable on the distance to the nearest town Primary Education for mothers (reference
Category same as in illiterate) 0.582 (0.031)* 0.528 (0.047)*
considered for measuring the conven-
Middle School for mothers (reference
tional diffusion argument turned out toCategory same as in illiterate) 0.331 (0.036)*
be insignificant. Muslim (1=Muslim; 0=others) 0.554 (0.026)* 0.375 (0.047)*
Thus the analysis demonstrated that it SC/ST (1=SC/ST; 0=others). 0.140 (0.017)*
Media Exposure (1=Yes; 0=No) -0.121 (0.014)* -0.162 (0.031)*
is important to distinguish different levels
Working mothers (1=yes; 0=No) -0.123(0.014)* -0.391 (0.04)*
while understanding the complex socialStandard of Living Index -0.004 (0.001)*
process such as fertility variation. A lotMember of a Self Help Group (1 =Yes; 0=No) - 0.060 (0.037)
remains to be understood at the commiu- Living in urban areas (1=Yes; 0=No) - -0.221 (0.056)*
Village Level Predictors
nity levels, as village level variation still
Women's Association (1=Yes; 0NNo) -0.099 (0.024)*
remains significant after taking into con-Number of Group meetings on Family
sideration the group interaction variables.Planning last one year -0.005 (0.001)*
Distance to the Nearest Town 0.001 (0.001)
Random Part
Conclusions
Constant -5.360 (0.087)* -3.886 (0.172)*
Level-3 Between District Variation - 0.038 (0.012)*
Level-2 Between PSU Variation 0.143 (0.007)* 0.055 (0.011)*
The recent fertility transition in many
Level-1 Between Women Variation 1.922 (0.012)* 1.875 (0.027)*
developing countries is increasingly get-
ting attention as it does not conform toFigures in the bracket are the standard errors. Media
Note: exposure is the expo
economic or social development argument.once a week or going to cinema at least once a month.
Standard of living index is calculated using asset holding and facilities at the h
Most of the conventional household level
of its estimation see National Family Health Survey (2000). However, final
theories become poor predictors of recent considered as such and did not classify into three like in NFHS-2 report.
fertility decline. Hence the explanations * Significant at 5 per cent level.

Economic and Political Weekly October 26, 2002 4453

This content downloaded from

14.139.114.34 on Fri, 19 Mar 2021 06:47:43 UTC
All use subject to https://about.jstor.org/terms
formed at the community level especially Census of India (1991): 'District Level Estimates Application to the Demographic Transition',
of those of small family size. This paper of Fertility and Child Mortality for 1991 and University of Pennsylvania (unpublished).
Their Inter Relations with Other Variables', National Family Health Survey (1995): India,
while emphasising the role of community Occasional Paper No 1 of 1997, Registrar 1992-93, International Institute for Population
interaction also cautions jumping into wild General of India, New Delhi. Sciences, Bombay.
conclusion. Probably this paper emphasises Census of India (2001): 'Provisional Population -(NFHS-2) (2000): India, 1998-99, International
more on the importance of community but Totals, India, Paper 1 of 2001', Registrar Institute for Population Sciences, Mumbai and
General of India, Delhi. ORC MACRO, Maryland.
more needs to be explored at that level thus
Cleland, J andC Wilson (1987): 'Demand TheoriesOrg-Marg Research Ltd (2000): 'Consultancy on
giving a new mandate for social research of the Fertility Transition: An Iconoclastic Strengthening of SeV Help Group in Andhra
from household to community. ll View', Population Studies, 41 (1), pp 5-50. Pradesh - SWOT Analysis of the Groups
Coleman, James S (1990): Foundations of Social Surveyed in Andhra Pradesh (Sponsored by
Address for correspondence: Theory, Cambridge, MA: Harvard University Department of Rural Development, Govern-
smdev@hd2.dot.net.in Press. ment of Andhra Pradesh, Hyderabad).
Demographic Health Survey (1997): Bangladesh, Pitt, Mark, M, Shahidur R Khandker, Signe-Mary
1996-97, National Institute of Population Re- Mckernan and M Abdul Latif (1999): 'Credit
[This paper is based on a large research study on
the causes of fertility decline in India and search and Training, Dhaka, Mitra Associates, Programmes for the Poor and Reproductive
Dhaka and Macro International, Maryland. Behaviour in Low Income Countries: Are the
Bangladesh sponsored by South Asia Network of
Entwisle, Barbara, John B Casterline and Hussein Reported Casual Relationships the Result of
Economic Research Institutes (SANEI), New Delhi.
A-A Sayed (1989): 'Villages as Contexts for Heterogeneity Bias', Demography, Vol 36,
We sincerely acknowledge the comments received
Contraceptive Behaviour in. Rural Egypt', (1), pp 1-21.
from the participants of two seminars at third
American Sociological Review, 54 (6), Ratcliffe, John. (1978): 'Social Justice and
annual conference of the SANEI at New Delhi,
pp 1019-34. Demographic Transition: Lessons from Kerala
August 28-31, 2001 and Centre for Economic and
Goldstein, H (1995): Multilevel Statistical Models State', International Journal of Health
Social Studies, Hyderabad.] 2nd edition, Arnold, London. Services 8, No 1:123-44.
Guilmoto, Christophe Z (2000): 'The Geography Retherford, R D (1985): A Theory of Marital
References of Fertility in India (1981-1991) 'in C S Fertility Transition', Population Studies,.
Guilmoto and A Vaguet (eds)', Essays on Vol 39 (2), pp 249-68.
Alexander, Jeffrey C (1988): Action and Its Population and Space in India, French Institute Sajeda, Amin, Alaka Malwade Basu and Rob
Environments: Towards a New Synthesis, Pondicherry, Pondicherry. Stephenson (2000): 'Spatial Variation in
Columbia University Press, New York. Guilmoto, Christophe Z and S Irudaya Rajan Contraceptive Use in Bangladesh: Looking
Anandhi, S (1996): 'Reproductive Bodies and (forthcoming): 'Half a Century of Fertility Beyond the Borders', Working Paper No 138,
Early 20th Century Tamil Nadu', Paper Transition in Indian Districts, 1951-2001', Policy Research Division, Population Council,
presented at 26th Interdisciplinary Research' Population and Development Review. New York.
Methodology Workshop on Rethinking IndianJames, K S (1999): 'Fertility Decline in Andhra Srinivasan, K (1995): Regulating Reproduction
Modernity: The Political Economy of Indian Pradesh: A Search for Alternative Hypo- in India's Population: Efforts, Results and
Sexuality, Madras Institute of Development theses', Economic and Political Weekly, Recommendations, Sage Publication, New
Research, Madras. 34 (8), pp 491-99. Delhi.
Antony, T V (1992): 'The Family Planning Krishnan,T N (1976): 'Demographic Transition in Srinivasan, K et al (1991): 'The Effect of Family
Programme: Lessons from Tamil Nadu's Kerala: Facts and Factors', Economic and Poli- Planning Components on Contraceptive
Experience', Paper presented at the symposium tical Weekly 11, No 31-33(August): 1203-24. Acceptance in Four Indian States', Inter-
on India's Development in the 1990s, CentreMontgomery, Mark, R and John B Casterline national Family Planning Perspectives,
for Policy Research, New Delhi. (1998): 'Social Networks and the Diffusion No 1:14-24.
Bangladesh Institute of Development Studies of Fertility Control', Working Paper No 119, Van De Kaa, D J (1996): 'Anchored Narratives:
(2001): 'Bangladesh Human Development Population Council, New York, New York. The Story and Findings of Half a Century of
Report 2000: Fighting Human Poverty', Munshi Kaivan and Jacques Myaux (2000): 'Social Research into the Determinants of Fertility',
Bangladesh Institute of Development Studies, Change and Individual Decision: With an Population Studies, 50, (3) pp 389-43.
Dhaka.
Basu, Alaka Malwade and Sajeda Amin (2000):
'Conditioning Factors for Fertility Decline in
Bengal: History, Language Identity and REVIEW OF AGRICULTURE
Openness to Innovations, Populatipn and
July 27, 2002
Development Review, 26 (4), pp 761-94.
Bhat, P N M (1996): 'Contours of Fertility Decline
Reforms in Agricultural Extension: New Policy Framework - Rita Sharma
in India: A District Level Study Based on the
1991 Census', Discussion paper series No 13, Crisis in Agrarian Economy in Punjab: Some Urgent Steps - H S Sidhu
UNDP Project, Centre for Development
Rice Production in Punjab: Systems, Varietal Diversity,
Studies, Thiruvananthapuram.
Bhat, P N M and S Irudaya Rajan (1990): Growth and Sustainability - Karam Singh,
'Demographic Transition in Kerala Revisited', Sajla Kalra
Economic and Political Weekly 25, (September
Agriculture in India and Pakistan, 1900-95:
1-8): 1957-80.
A Further Note - Takashi Kurosaki
Bhat, P N M and Francis Savier (1999): 'Findings
of National Family Health Survey: Regional Sagging Agricultural Commodity Exchan
Analysis, Economic and Political Weekly,
October, 16-23, pp 3008-32.
Constraints and Revival Policy Options - K G Sahadevan
Bocquet-Appel, Pean Pierre and Lucienne Jakobi Indian Agricultural Commodity Futures Markets:
(1998): 'Evidence for a Spatial Diffusion of A Performance Survey - Gopal Naik,Sudhir Kumar Jain
Contraception at the Onset of the Fertility
Transition in Victorian Briton, Population, For copies write to
(Special issue), pp 181-204.
Circulation Manager
Bongaarts, John and Susan Cotts Watkins (1996):
Social Interactions and Contemporary Fertility Economic and Political Weekly
Transition, Population and Development Hitkari House, 284, Shahid Bhagatsingh Road, Mumbai 400 001
Review, 22 (4): 639-82.

4454 Economic and Political Weekly October 26, 2002

This content downloaded from

14.139.114.34 on Fri, 19 Mar 2021 06:47:43 UTC
All use subject to https://about.jstor.org/terms