02-06-2005, 01:39 AM
EPW 1-29-2005
Role of Religion in Fertility Decline - Part 1
P N MARI BHAT, A J FRANCIS ZAVIER
****After a lot of mathematical calculations they prove that muslim excess fertility has nothing to do with education or poverty, rather it is due to deliberate over-breeding
The Case of Indian Muslims
The fertility of Muslims, which was about 10 per cent higher than that of Hindus before
independence, is now 25 to 30 per cent higher than the Hindu rate, and the difference
according to religion is larger than the difference between the forward and depressed Hindu
castes and tribes.
This paper subjects the micro data from the National Family Health
Surveys to a multivariate analysis to assess the contribution of socio-economic factors to the
fertility differential by religion. It also explores the possible reasons for the large,
residual effect of religion on fertility, and causes for the religious disparities in socio-economic
conditions. The paper concludes with an assessment of the implications of the current
demographic trends for the future population sizes of the two religious groups.
P N MARI BHAT, A J FRANCIS ZAVIER
***Kingsley Davis, a noted sociologist and a renowned demographer,
was one of the first to comment upon the Hindu-Muslim
fertility difference. In his monumental work on the population
of India and Pakistan, he noted that child-woman ratios of
Muslims in undivided India were about 12-14 per cent higher
than among Hindus [Davis 1951].2 As few in India used contraception
then, the Hindu-Muslim fertility difference of the time
must have arisen from differences in supply-side factors (i e, biosocial
constraints on fertility) such as marriage and remarriage,
abstinence for religious reasons and breastfeeding. Among them,
differences in widow remarriage and duration of sexual abstinence
after birth could have been particularly important.
***There is widespread belief, especially among the âHindu Rightâ,
that the practice of polygyny is the main reason for the higher
fertility of Muslims. But since only women give birth to children,
polygyny can influence fertility of a population only by raising
the proportion of married women.3 But the available data show
that Muslim women on an average did not marry at a significantly
earlier age than Hindu women, and those remaining never married
were extremely few in both the communities. Nevertheless, the
data for pre-partition India did show that the proportion of
currently married women was higher among Muslims than Hindus
by about 4 per cent [Davis 1951]. This excess was largely due
to the higher frequency of widow remarriages among Muslims.
Thus the extent of polygyny among Muslims was only such that
it balanced the excess supply of married women that resulted
from widow remarriage. Consequently, the fertility enhancing
effects of the two factors were not mutually exclusive. Together
they accounted for about one-third of the excess child-woman
ratios of Muslims.
***fertility of Muslims
is even higher than those of the bottom-most groups in the Hindu
caste hierarchy, namely, the scheduled castes (SC) and scheduled
tribes (ST). For example, the estimates of TFR from NFHS-2
show that in comparison with the average fertility of Hindus,
the fertility of SCs is 13 per cent higher, that of STs is 10 per
cent higher, and of other backward castes (OBC) is only 2 per
cent higher. But as already been noted, Muslim fertility is 29
per cent higher than the average Hindu fertility. Similar differentials
are indicated by the data from NFHS-1, and that on
completed family size (Table 1). Therefore, there is little ground
for rationalising the higher fertility of Muslims by pointing out
the presence of equally high levels of fertility among some Hindu
groups.
****But among the younger cohorts there
has been a steady rise in the religious fertility difference. In the
most recent cohorts reaching the end of their reproductive span,
the difference is about one birth per woman, or an excess of about
25 per cent among Muslim women. The rise in Hindu-Muslim
fertility difference has been particularly sharp in urban areas
where the recent cohorts show a difference of 1.2 children, or
an excess of more than 30 per cent in Muslim fertility over that
of the Hindus.
It is also sometimes contended that higher fertility of Muslims
could be because they are concentrated in areas where fertility
is generally high, such as in northern India [e g, Jeffery and Jeffery
2000, 2002]. But a careful examination of the available data
shows that in every major state, fertility is higher among Muslims
than Hindus, and the gap between the two has been widening
with time. Data from both census and surveys substantiate this
conclusion. The data released recently from the 2001 Census
show that child-woman ratio (CWR, defined as the ratio of
children less than seven years to females aged seven years and
over) is higher among Muslims than Hindus in all the states except
Chhattisgarh and Madhya Pradesh (Table 3). Even the two
anomalous cases can be explained by the fact that more than 60
per cent of Muslims in the two states live in urban areas compared
to less than 25 per cent of the Hindus. It may also be noted that
the child-woman ratio of Muslims is higher by 24 per cent than
that of Hindus at the all-India level, whereas in Kerala and West
Bengal it is higher by 43 and 59 per cent, respectively.
***It is occasionally asserted that fertility of Muslims in a state
such as Kerala is lower than fertility of Hindus in a state like
Uttar Pradesh, and hence it cannot be claimed that Muslims have
higher fertility than Hindus. But such an argument is absurd
because factors other than religion also have an influence on
fertility. Data from the NFHS also show that fertility of illiterate
women in Kerala is lower than fertility of women who have
completed high school in Uttar Pradesh. Could this evidence be
used to refute the effect of female education on fertility?
****It is the emergence of the religious difference in contraceptive
practice since the 1970s that is chiefly responsible for the widened
gulf between Hindu-Muslim fertility. The data on contraceptive
use among Muslims and Hindus from five national surveys are
shown in Table 5. The first survey conduced by the Operations
Research Group in 1970 showed a difference of 5 per cent in
the current use of contraception among married women between
Hindus and Muslims at the national level (14 and 9 per cent,
respectively). The second survey conducted in 1980 disclosed
that this difference increased to 12 per cent. Subsequent surveys,
including the NFHS, have showed that this difference has remained
more or less the same. The contribution of the difference
in contraceptive practice to the difference in fertility between
Hindus and Muslims can be assessed as the ratio of women not
using contraception in the two communities.8 The computations
made on this basis show that the difference in contraceptive use
alone would have caused fertility of Muslims to be higher than
that of Hindus by 6 per cent in 1970, 23 per cent in 1980, and
28 per cent in 1998-99 (Table 5).
***It is often argued that much of the difference in fertility between
Hindus and Muslims is because of higher levels of poverty and
illiteracy among Muslims. But few have examined carefully the
available data on the socio-economic characteristics of the
population and the extent to which religion itself could have
contributed to the emergence of such differences. For explaining
the differentials in fertility, the Hindu-Muslim differences in
(i) rural-urban distribution of the population, (ii) literacy and
educational levels, (iii) income and poverty and (iv) female
autonomy and son preference are of particular interest. We shall
examine the religious differentials in each of these in some detail.
***population and the extent to which religion itself could have
contributed to the emergence of such differences. For explaining
the differentials in fertility, the Hindu-Muslim differences in
(i) rural-urban distribution of the population, (ii) literacy and
educational levels, (iii) income and poverty and (iv) female
autonomy and son preference are of particular interest. We shall
examine the religious differentials in each of these in some detail.
***According to this source, the difference
between Hindus and Muslims is 3 percentage points in female
literacy (53 and 50 per cent) and 8 percentage points in male
literacy (76 and 68 per cent
****A probable reason is that concerns about the
secularising influence of modern education is prompting the
followers of more organised religion like Islam to keep away
their children from public schools. But some Muslim scholars
and western observers contend that discrimination in recruitment
to public offices is the main source for the lack of interest of
Muslims in acquiring higher education [e g, Hasan 1997; Jeffery
and Jeffery 2000; Harris-White 2003]. Statistics have been compiled
to show that Muslims form only about 3 per cent of those
employed in the Indian administrative service, police, railways
and nationalised banks. The figures quoted do not suggest a high
level of discrimination given that they form only about 5 per
cent of those graduated from college, and that there could be
differences in the quality of education they acquire.
***As the figure shows, although there is an inverse
relationship between fertility and literacy in both the communities,
for the same level of female literacy CWRs of Muslims
are generally higher than those of Hindus.
****According to these regressions, for the CWRs of Muslims and
Hindus to be equal, the level of female literacy among the former
has to be higher by about 15 percentage points. Although the
fitted lines suggest that with the increase in female literacy
Muslim fertility falls more rapidly than Hindu fertility, the fertility
differential does not completely disappear even when literacy
attains 100 per cent (Figure 5). As a matter fact, the fitted lines
suggest that the percentage difference in fertility between the two
communities increases with female literacy. For example, when
female literacy is 20 per cent, the predicted CWR for Muslims
is higher than that of Hindus by 15 per cent, whereas when it
is 80 per cent, the predicted CWR of Muslims is higher than
that of Hindus by 21 per cent. Thus the bivariate analysis of the
macro data from the census suggests that fertility of Muslims
is higher than that of Hindus, irrespective of their literacy levels.
***As the
table shows, in rural areas, there is hardly any difference in the
poverty ratios of Muslims and Hindus. As per the data from the
43rd round, 41 per cent of Hindus and 40 per cent of Muslims
were living under poverty.
***A recent study of several Asian
countries that included India found that âgreater pronatalist
behaviour/attitude cannot be accounted for by any general tendency
of Muslim women to have less autonomy than non-Muslim
womenâ [Morgan et al 2002:533].
***tables. The estimated
effects of education, exposure to mass media, caste/tribe and
duration of marriage are significant and show expected signs in
almost all regressions. The dummy variable representing the
survey round (NFHS-2) shows strong effects in all the regressions,
indicating that changes in socio-economic variables included
in the regressions cannot explain all the changes that
occurred in the fertility measures during the inter-survey period.
We have discussed implications of these results at length elsewhere
[Bhat and Zavier 2004]. Here we shall focus only on the
measured effects of religion.
Even after controlling for all the known socio-economic
factors, the dummy variable representing Muslims is strongly
significant in all the regressions, both for rural and urban areas.
Its effect is negative on contraceptive use, and positive in all
other regressions. It is to be noted that because we have controlled
for scheduled caste/tribes in the regressions, the dummy variable
for Muslims captures the difference between Muslims and Hindus
belonging to castes other than SC and ST (caste Hindus, hereafter).
But as the coefficients of dummy variables for SC and
ST show, the differences between caste Hindus and SC/ST are
smaller than the difference between caste Hindus and Muslims.
This is true for all regressions except that of ideal family size
regression for urban areas.
***Does the Muslim effect disappear at higher levels of education?
The education-Muslim interaction effects are not statistically
significant in most regressions. In some regressions they are
even in the same direction as the main effect of being Muslim
(e g, in ideal family size regressions), indicating that the differ-
ence between Muslims and Hindus, rather than reducing gets
widened at higher levels of education. Only the cohort fertility
regression for urban areas shows diminishing effects at higher
educational levels. But even in this case, the higher fertility of
Muslims does not completely disappear at higher educational
levels
***Finally, we assess the residual contribution of religion. Since
the dummy variable for Muslims provides comparison only with
caste Hindus, to make comparison with an average Hindu we
should combine the effects of religion and caste. There also is
the question of where to classify the interaction effect of education
and religion. There could be arguments both for and against
its inclusion under the religious effect. If included with religion,
86-103 per cent of the observed Hindu-Muslim difference in
fertility indicators in rural areas, and 55-77 per cent of the
difference in urban areas, would be due to religion.17 If the
interaction effect were omitted, 75-100 per cent of the total
difference in rural areas, and 56-99 per cent of the total difference
in urban areas, would be due to religion. If the contribution of
duration of marriage were to be included (since it measures the
effect of differences in past fertility), the contribution attributed
to religion would be even larger. To conclude, even if it is assumed
that religious differences in socio-economic factors were produced
exogenously, they cannot explain no more than one-fourth
of the Hindu-Muslim differences in fertility measures in rural
areas, and half of the differences in urban areas. If religion were
the cause for some of the differences in socio-economic factors,
then the independent contribution of theses factors to the religious
fertility differential would be even smaller.
Role of Religion in Fertility Decline - Part 1
P N MARI BHAT, A J FRANCIS ZAVIER
****After a lot of mathematical calculations they prove that muslim excess fertility has nothing to do with education or poverty, rather it is due to deliberate over-breeding
The Case of Indian Muslims
The fertility of Muslims, which was about 10 per cent higher than that of Hindus before
independence, is now 25 to 30 per cent higher than the Hindu rate, and the difference
according to religion is larger than the difference between the forward and depressed Hindu
castes and tribes.
This paper subjects the micro data from the National Family Health
Surveys to a multivariate analysis to assess the contribution of socio-economic factors to the
fertility differential by religion. It also explores the possible reasons for the large,
residual effect of religion on fertility, and causes for the religious disparities in socio-economic
conditions. The paper concludes with an assessment of the implications of the current
demographic trends for the future population sizes of the two religious groups.
P N MARI BHAT, A J FRANCIS ZAVIER
***Kingsley Davis, a noted sociologist and a renowned demographer,
was one of the first to comment upon the Hindu-Muslim
fertility difference. In his monumental work on the population
of India and Pakistan, he noted that child-woman ratios of
Muslims in undivided India were about 12-14 per cent higher
than among Hindus [Davis 1951].2 As few in India used contraception
then, the Hindu-Muslim fertility difference of the time
must have arisen from differences in supply-side factors (i e, biosocial
constraints on fertility) such as marriage and remarriage,
abstinence for religious reasons and breastfeeding. Among them,
differences in widow remarriage and duration of sexual abstinence
after birth could have been particularly important.
***There is widespread belief, especially among the âHindu Rightâ,
that the practice of polygyny is the main reason for the higher
fertility of Muslims. But since only women give birth to children,
polygyny can influence fertility of a population only by raising
the proportion of married women.3 But the available data show
that Muslim women on an average did not marry at a significantly
earlier age than Hindu women, and those remaining never married
were extremely few in both the communities. Nevertheless, the
data for pre-partition India did show that the proportion of
currently married women was higher among Muslims than Hindus
by about 4 per cent [Davis 1951]. This excess was largely due
to the higher frequency of widow remarriages among Muslims.
Thus the extent of polygyny among Muslims was only such that
it balanced the excess supply of married women that resulted
from widow remarriage. Consequently, the fertility enhancing
effects of the two factors were not mutually exclusive. Together
they accounted for about one-third of the excess child-woman
ratios of Muslims.
***fertility of Muslims
is even higher than those of the bottom-most groups in the Hindu
caste hierarchy, namely, the scheduled castes (SC) and scheduled
tribes (ST). For example, the estimates of TFR from NFHS-2
show that in comparison with the average fertility of Hindus,
the fertility of SCs is 13 per cent higher, that of STs is 10 per
cent higher, and of other backward castes (OBC) is only 2 per
cent higher. But as already been noted, Muslim fertility is 29
per cent higher than the average Hindu fertility. Similar differentials
are indicated by the data from NFHS-1, and that on
completed family size (Table 1). Therefore, there is little ground
for rationalising the higher fertility of Muslims by pointing out
the presence of equally high levels of fertility among some Hindu
groups.
****But among the younger cohorts there
has been a steady rise in the religious fertility difference. In the
most recent cohorts reaching the end of their reproductive span,
the difference is about one birth per woman, or an excess of about
25 per cent among Muslim women. The rise in Hindu-Muslim
fertility difference has been particularly sharp in urban areas
where the recent cohorts show a difference of 1.2 children, or
an excess of more than 30 per cent in Muslim fertility over that
of the Hindus.
It is also sometimes contended that higher fertility of Muslims
could be because they are concentrated in areas where fertility
is generally high, such as in northern India [e g, Jeffery and Jeffery
2000, 2002]. But a careful examination of the available data
shows that in every major state, fertility is higher among Muslims
than Hindus, and the gap between the two has been widening
with time. Data from both census and surveys substantiate this
conclusion. The data released recently from the 2001 Census
show that child-woman ratio (CWR, defined as the ratio of
children less than seven years to females aged seven years and
over) is higher among Muslims than Hindus in all the states except
Chhattisgarh and Madhya Pradesh (Table 3). Even the two
anomalous cases can be explained by the fact that more than 60
per cent of Muslims in the two states live in urban areas compared
to less than 25 per cent of the Hindus. It may also be noted that
the child-woman ratio of Muslims is higher by 24 per cent than
that of Hindus at the all-India level, whereas in Kerala and West
Bengal it is higher by 43 and 59 per cent, respectively.
***It is occasionally asserted that fertility of Muslims in a state
such as Kerala is lower than fertility of Hindus in a state like
Uttar Pradesh, and hence it cannot be claimed that Muslims have
higher fertility than Hindus. But such an argument is absurd
because factors other than religion also have an influence on
fertility. Data from the NFHS also show that fertility of illiterate
women in Kerala is lower than fertility of women who have
completed high school in Uttar Pradesh. Could this evidence be
used to refute the effect of female education on fertility?
****It is the emergence of the religious difference in contraceptive
practice since the 1970s that is chiefly responsible for the widened
gulf between Hindu-Muslim fertility. The data on contraceptive
use among Muslims and Hindus from five national surveys are
shown in Table 5. The first survey conduced by the Operations
Research Group in 1970 showed a difference of 5 per cent in
the current use of contraception among married women between
Hindus and Muslims at the national level (14 and 9 per cent,
respectively). The second survey conducted in 1980 disclosed
that this difference increased to 12 per cent. Subsequent surveys,
including the NFHS, have showed that this difference has remained
more or less the same. The contribution of the difference
in contraceptive practice to the difference in fertility between
Hindus and Muslims can be assessed as the ratio of women not
using contraception in the two communities.8 The computations
made on this basis show that the difference in contraceptive use
alone would have caused fertility of Muslims to be higher than
that of Hindus by 6 per cent in 1970, 23 per cent in 1980, and
28 per cent in 1998-99 (Table 5).
***It is often argued that much of the difference in fertility between
Hindus and Muslims is because of higher levels of poverty and
illiteracy among Muslims. But few have examined carefully the
available data on the socio-economic characteristics of the
population and the extent to which religion itself could have
contributed to the emergence of such differences. For explaining
the differentials in fertility, the Hindu-Muslim differences in
(i) rural-urban distribution of the population, (ii) literacy and
educational levels, (iii) income and poverty and (iv) female
autonomy and son preference are of particular interest. We shall
examine the religious differentials in each of these in some detail.
***population and the extent to which religion itself could have
contributed to the emergence of such differences. For explaining
the differentials in fertility, the Hindu-Muslim differences in
(i) rural-urban distribution of the population, (ii) literacy and
educational levels, (iii) income and poverty and (iv) female
autonomy and son preference are of particular interest. We shall
examine the religious differentials in each of these in some detail.
***According to this source, the difference
between Hindus and Muslims is 3 percentage points in female
literacy (53 and 50 per cent) and 8 percentage points in male
literacy (76 and 68 per cent
****A probable reason is that concerns about the
secularising influence of modern education is prompting the
followers of more organised religion like Islam to keep away
their children from public schools. But some Muslim scholars
and western observers contend that discrimination in recruitment
to public offices is the main source for the lack of interest of
Muslims in acquiring higher education [e g, Hasan 1997; Jeffery
and Jeffery 2000; Harris-White 2003]. Statistics have been compiled
to show that Muslims form only about 3 per cent of those
employed in the Indian administrative service, police, railways
and nationalised banks. The figures quoted do not suggest a high
level of discrimination given that they form only about 5 per
cent of those graduated from college, and that there could be
differences in the quality of education they acquire.
***As the figure shows, although there is an inverse
relationship between fertility and literacy in both the communities,
for the same level of female literacy CWRs of Muslims
are generally higher than those of Hindus.
****According to these regressions, for the CWRs of Muslims and
Hindus to be equal, the level of female literacy among the former
has to be higher by about 15 percentage points. Although the
fitted lines suggest that with the increase in female literacy
Muslim fertility falls more rapidly than Hindu fertility, the fertility
differential does not completely disappear even when literacy
attains 100 per cent (Figure 5). As a matter fact, the fitted lines
suggest that the percentage difference in fertility between the two
communities increases with female literacy. For example, when
female literacy is 20 per cent, the predicted CWR for Muslims
is higher than that of Hindus by 15 per cent, whereas when it
is 80 per cent, the predicted CWR of Muslims is higher than
that of Hindus by 21 per cent. Thus the bivariate analysis of the
macro data from the census suggests that fertility of Muslims
is higher than that of Hindus, irrespective of their literacy levels.
***As the
table shows, in rural areas, there is hardly any difference in the
poverty ratios of Muslims and Hindus. As per the data from the
43rd round, 41 per cent of Hindus and 40 per cent of Muslims
were living under poverty.
***A recent study of several Asian
countries that included India found that âgreater pronatalist
behaviour/attitude cannot be accounted for by any general tendency
of Muslim women to have less autonomy than non-Muslim
womenâ [Morgan et al 2002:533].
***tables. The estimated
effects of education, exposure to mass media, caste/tribe and
duration of marriage are significant and show expected signs in
almost all regressions. The dummy variable representing the
survey round (NFHS-2) shows strong effects in all the regressions,
indicating that changes in socio-economic variables included
in the regressions cannot explain all the changes that
occurred in the fertility measures during the inter-survey period.
We have discussed implications of these results at length elsewhere
[Bhat and Zavier 2004]. Here we shall focus only on the
measured effects of religion.
Even after controlling for all the known socio-economic
factors, the dummy variable representing Muslims is strongly
significant in all the regressions, both for rural and urban areas.
Its effect is negative on contraceptive use, and positive in all
other regressions. It is to be noted that because we have controlled
for scheduled caste/tribes in the regressions, the dummy variable
for Muslims captures the difference between Muslims and Hindus
belonging to castes other than SC and ST (caste Hindus, hereafter).
But as the coefficients of dummy variables for SC and
ST show, the differences between caste Hindus and SC/ST are
smaller than the difference between caste Hindus and Muslims.
This is true for all regressions except that of ideal family size
regression for urban areas.
***Does the Muslim effect disappear at higher levels of education?
The education-Muslim interaction effects are not statistically
significant in most regressions. In some regressions they are
even in the same direction as the main effect of being Muslim
(e g, in ideal family size regressions), indicating that the differ-
ence between Muslims and Hindus, rather than reducing gets
widened at higher levels of education. Only the cohort fertility
regression for urban areas shows diminishing effects at higher
educational levels. But even in this case, the higher fertility of
Muslims does not completely disappear at higher educational
levels
***Finally, we assess the residual contribution of religion. Since
the dummy variable for Muslims provides comparison only with
caste Hindus, to make comparison with an average Hindu we
should combine the effects of religion and caste. There also is
the question of where to classify the interaction effect of education
and religion. There could be arguments both for and against
its inclusion under the religious effect. If included with religion,
86-103 per cent of the observed Hindu-Muslim difference in
fertility indicators in rural areas, and 55-77 per cent of the
difference in urban areas, would be due to religion.17 If the
interaction effect were omitted, 75-100 per cent of the total
difference in rural areas, and 56-99 per cent of the total difference
in urban areas, would be due to religion. If the contribution of
duration of marriage were to be included (since it measures the
effect of differences in past fertility), the contribution attributed
to religion would be even larger. To conclude, even if it is assumed
that religious differences in socio-economic factors were produced
exogenously, they cannot explain no more than one-fourth
of the Hindu-Muslim differences in fertility measures in rural
areas, and half of the differences in urban areas. If religion were
the cause for some of the differences in socio-economic factors,
then the independent contribution of theses factors to the religious
fertility differential would be even smaller.