ISSN (Print) - 0012-9976 | ISSN (Online) - 2349-8846

A+| A| A-

Amartya Sen and Identities


Trivialising Religion

Causes of Higher Muslim Fertility


he collection of papers in the EPW special issue on fertility behaviour of Hindus and Muslims (January 29, 2005) has thrown up several contentious issues that need to be debated further. Although a consensus seems to have emerged that the current Hindu-Muslim fertility differential cannot wholly be explained by socioeconomic or residential differences, there remains considerable disagreement on the size and trends in the fertility differential, and more so on its causes. As an author of one of the papers included in the special issue [Bhat and Zavier 2005], one last attempt is made here to explain why I disagree with the positions taken in some of the papers.

Some authors are of the view that the current difference in fertility between Hindus and Muslims is small but its scale is usually exaggerated in the media [Bhagat and Praharaj 2005; Jeffrey and Jeffrey 2005]. When Jeffrey and Jeffrey make such a claim, the reference invariably is towards propaganda materials of the Hindu nationalists. But due to their ideological leaning, Hindu nationalists would be prone to exaggeration, and thus are not the proper basis for objective assessment. In scientific works, the significance of a differential is judged in relation to others, or from its consequences. Decades of demographic research in India have shown that among the many fertility differentials studied, the Hindu-Muslim differential is one of the most enduring ones. For example, Jack Caldwell, whose micro research in south India gets strong endorsement from Jeffrey and Jeffrey, had this to say on this issue: “Perhaps because of greater pressure on the poor, especially during the Emergency, there are no significant differentials in fertility or family planning practice by socio-economic status, caste grouping, or education. There is only one important differential, and that is by religion” [Caldwell et al 1982:714]. An element of exaggeration in this statement cannot be ruled out, as no hard data on the differentials were presented. But it is unlikely that a keen observer like Caldwell would have missed much. As per NFHS-2, the difference in TFR between illiterate women and those who completed high school is 1.5 births (3.5 and 2.0, respectively). A difference of one birth per woman between Hindus and Muslims is thus equivalent to the effect of about six years of schooling.1 A decrease of only one birth per woman is needed for India to achieve the replacement level fertility. Thus a difference of one birth per woman, or roughly the order of 30 per cent, should not be dismissed as insignificant.

It is also contended that the current difference in fertility rates of Hindus and Muslims is only temporary, and soon the levels of the two would converge. But this also applies to fertility differentials by education or income. It is also likely that soon, the fertility difference between Uttar Pradesh and Kerala would disappear. So, should we be ignoring the difference by educational levels, or that between north and south India? Trivialising the religious difference is in nobody’s interest because the Hindu nationalists, who need to be convinced, would dismiss it as minority appeasement and among Muslims, it could diminish the desire for introspection.

Explaining Fertility Differential

As causes of the fertility differential, I find at least three separate lines of argument that studiously avoid implicating religion. First, the higher fertility of Muslims is attributed to their minority status, and constant threats and vilification campaign of Hindu nationalists [Bhagat and Praharaj 2005; Dharmalingam et al 2005; Jeffrey and Jeffrey 2005]. Second, it is attributed to their higher natural fertility and delayed onset of fertility decline, owing to their lower socioeconomic status [James and Nair 2005; Krishnaji and James 2005]. Third, Borooah and Iyer (2005) attribute it to the lower degree of daughter aversion among Muslims. The last one, while invoking religion, highlights a positive attribute of Muslims. I shall take up these propositions in the reverse order, and argue why they lack substance.

First, it should be noted that demographers distinguish between sex preference and preference for number of children, whereas Borooah and Iyer (B&I) do not. For B&I, son preference means desiring more sons relative to that in a reference group, and daughter aversion means desiring fewer daughters than the same group. As demographers control for the effect of preference for total number of children while measuring the effect of sex preference on fertility, for them son preference and daughter aversion are one and the same. But in B&I’s terminology, they are not necessarily complementary. Besides this conceptual difference, the analysis presented by B&I in support of their daughter aversion hypothesis has a technical flaw.2 The results based on a partial sample of women who have terminated their fertility could be biased because the probability of selection into the sample is not the same for Hindus and Muslims. As non-use of contraception could also be due to sex preference, it is necessary to consider the entire sample of women in totality. Also, considering only terminal method users, as B&I do, would be a mistake in this case because religious injunctions against sterilisation could be forcing many Muslims to use temporary methods for terminating fertility.

A method devised by Arnold (1985) gets around this problem, as it uses the entire sample of women and the data on contraceptive use status by number and sex of living children. The method measures the effect of sex composition of children on contraceptive use by controlling for the effect of total number of children. As noted in our paper, this method applied to the NFHS data, showed that sex preference depressed contraceptive use among both Hindus and Muslims, but more so in the former. If one wishes to distinguish the effects of son preference from daughter aversion, the proper thing to do is to regress contraceptive use on number of sons and number of daughters. Table 1 shows the results of such a regression

Economic and Political Weekly March 26, 2005

done using NFHS-2 data for Hindus and Muslims. The estimated odds ratios show that the odds of using contraception increase with the number of sons more rapidly in Hindus than in Muslims. That is, Muslims end up with more sons than Hindus before accepting contraception. Although the estimated rise is more muted, the same pattern is found in the case of daughters. In other words, compared to Hindus, Muslims desire more sons as well as daughters. Thus, in B&I’s terminology, fertility is higher among Muslims both on account of higher son preference and lower daughter aversion. But in demographer’s terminology, son preference (or daughter aversion) is higher among Hindus than Muslims. This is because the estimated odds ratios show that contraceptive use is determined more by the number of sons than by the number of daughters, and the sex difference in the odds ratio is higher for Hindus than for Muslims.

The argument of James et al has two interconnected parts: (i) Muslims have significantly higher natural fertility level that partly explains their higher fertility, and (ii) because of their lower socioeconomic status, Muslims were not exposed to the fertility declines that occurred before the 1980s; but since then, the pace of fertility decline among them has been comparable to that of Hindus. On the first, James and Nair were unable to find confirmatory evidence, as their analysis of proximate determinants showed that total fecundity rate (a measure of biological upper limit to fertility) is lower among Muslims than Hindus. The second part of their argument too rests on shaky evidence. As per the 1984 SRS survey, the Muslim TFR was 5.6 births per woman. But the available evidence suggests that the Muslim TFR in the 1960s was more than 6.5. Clearly, significant fertility decline had occurred among Muslims before the 1980s. The pace of fertility decline during 1970-78 was, if not faster, comparable to that which occurred in the late 1980s. It is known that mass vasectomy camps conducted during and before the Emergency period played a critical part in it. It would be naive to assume that such sterilisation drives did not affect the poor (vide, the above quote from Caldwell).

The contention of James et al that Muslims have kept pace with the Hindus since the ‘accelerated’ decline in fertility began in the 1980s is also fallible. The census data on Muslim-Hindu growth rates show that the growth difference between the two increased from 0.5 per cent per annum during 1971-81 to 0.8 per cent during 1981-91. The growth difference of 1981-91 was the highest ever recorded since the census exercises began tracking this trend. Thus the birth rate difference between the two communities must have widened during the 1980s. The problem with the data presented by James et al is that the estimates of TFR they use are derived from sample surveys that did not over-sample Muslims. Such estimates provide only a rough magnitude of the religious differential in fertility for the survey date, and are highly inadequate for inferring trends in fertility differentials. There are reasons to believe that the 1984 survey had exaggerated the Muslim-Hindu fertility differential while NFHS-2 in 1998-99 had underestimated it. As per the 1984 survey, the Muslim TFR was 24 per cent higher than that of Hindus. There also was a SRS survey in 1979, which showed that the excess Muslim fertility was only of the order of 12 per cent. By all accounts, fertility levels stagnated during 1978-83 because of excesses during the Emergency. How could then the religious difference change? The only possible explanation is sampling and non-sampling errors in the two surveys. Therefore, rather than using the estimates of only one survey, if an average is taken, Muslim fertility at the beginning of 1980s exceeded the Hindu fertility by about 18 per cent. NFHS-1 conducted in 1992-93 showed that the religious difference in current fertility had increased to 34 per cent, which is consistent with the trends in census growth rates.

Understanding Survey Methods

The estimates of TFR from NFHS-2 of 1998-99 showed that the excess Muslim fertility had declined to 29 per cent. But there is evidence to suggest that NFHS-2 had considerably underestimated fertility levels in Uttar Pradesh and Assam [Retherford and Mishra 2001]. As these two states have a strong presence of Muslims, the underestimation would have influenced TFR of Muslims more than that of Hindus. Another evidence of underestimation of Muslim TFR in NFHS-2 comes from the analysis of proximate determinants made by James and Nair (2005). They find that the total fecundity rate of Muslims is 6 per cent lower than that of Hindus (10.8 and 11.5, respectively). As total fecundity rate is derived from TFR by eliminating the fertility depressing influence of proximate determinants, it would carry the same bias as the TFR.3 For the total fecundity rate of Muslims and Hindus to be equal, TFR of Muslims would have to be raised by 6 per cent vis-à-vis that of Hindus. This would make the TFR of Muslims to be 37 per cent higher than that of Hindus. Thus since the beginning of 1980s, the excess fertility of Muslims may have increased from 18 per cent to 37 per cent. This gives quite a different picture from the one painted by James et al.Further, as per their own data on TFR by standard of living index [Krishnaji and James 2005], even among the poor, Muslims lagged behind Hindus in fertility decline during this period. In the low standard of living households, as per NFHS-1, the Muslim TFR was 33 per cent higher than the Hindu TFR, and as per NFHS-2, the difference was 29 per cent. Thus the contention that because of their lower socio-economic status Muslims were unaffected by earlier changes is too contrived.

In conclusion, the Hindu-Muslim fertility difference almost certainly increased in the 1980s. But in the 1990s, there was no further widening of the difference at the all-India level because the census growth difference during 1991-2001 was about the same as that during 1981-91, and NFHS-1 and NFHS-2 have shown more or less the same fertility difference between the two communities. However, this should not be interpreted as showing that fertility decline among Muslims accelerated in the 1990s. The reason for the unchanging religious fertility differential at the national level is most probably because, fertility decline among Hindus slowed down in several south Indian states as their fertility was nearing replacement level. The narrowing differential in southern states offset the widening differential in other states. This however cannot be checked from the NFHS data because sample sizes for Muslims are inadequate

Table 1: Results of Logistic Regressionof Current Use of Contraception onNumber of Sons and Daughters forHindus and Muslims, NFHS-2, All-India

Explanatory Hindus Muslims Variables Odds SE Odds SE Ratio Ratio

Number of sons

alive 1.831 0.015 1.238 0.030 Number of

daughters alive 1.194 0.008 1.069 0.021 Constant 0.373 0.006 0.389 0.023 N 66,136 10,121 Pseudo R2 0.136 0.034

Economic and Political Weekly March 26, 2005 to ascertain trends in the differentials. The census growth rates can throw some light on this, but because of their sensitivity to migration, they have to be interpreted carefully. Table 2 shows the average annual growth rates of population for Hindus and Muslims during 1981-91 and 1991-01. The declines in the growth rate between 1981-91 and 1991-01 were significantly more among Muslims than Hindus in Kerala, Tamil Nadu, Andhra Pradesh, West Bengal, Uttar Pradesh and Rajasthan. Amongst these, in the first four states, the growth rates of Hindus have fallen below

1.4 per cent per annum, indicating the presence of low levels of Hindu fertility. The larger decline of Muslim growth rate in West Bengal could also be because of the decrease in the flow of illegal migrants from Bangladesh, or their migration to other parts of India. In Uttar Pradesh and Rajasthan, the growth rates of Hindus have hardly fallen, but that of Muslims have shown large declines. This is unlikely due to significant falls in their fertility. The Muslim populations in Delhi, Haryana and Punjab have shown very high growth rates (4-6 per cent per annum), suggesting that the Muslims of Uttar Pradesh and Rajasthan are migrating to these states. Thus the trends in the census growth rates do suggest that the slowing down of fertility decline among Hindus in southern states helped Muslims to be on par with Hindus at the national level.

‘Minority Hypothesis’

Let us now examine critically the most popular of the explanations, namely, the minority hypothesis. Within this, I find three variants. First, it is argued that because they are a minority religion, Muslims are denied basic services, schooling and employment in post-independence India. The resulting social exclusion is responsible for the higher fertility of Muslims. Second, Hindu nationalists spread canards about unbridled fertility of Muslims, rampant polygynous unions and Muslims becoming a majority community within India. This tends to evoke a defensive response from the Muslims that keeps their fertility high. Third, communal riots and vilification campaigns make Muslims insecure, and desire more children for self-protection.

On the first, several papers included in EPW ’s special issue have shown that the lower socio-economic position of Muslims can explain only a small portion of their higher fertility. Therefore, for explaining higher fertility of Muslims, it hardly matters how the question of their social exclusion arose, whether from their minority status or from religious attitudes. Also, the evidence of lower mortality among Muslims indicates that the depiction of Muslims as a socially marginalised group is a bit overdone. On the second, those exposed to the propaganda of the Hindu Right, whether through print media or public meetings, would predominantly be Hindus rather than Muslims. Therefore, if at all such propaganda has a direct impact on fertility, it would be to increase Hindu fertility. However, if the argument is that such propaganda increases violence against Muslims, then it is undistinguishable from the insecurity syndrome examined below.

The argument that the sense of insecurity could lead to higher fertility of a minority group seems plausible at first sight. However, any such reaction necessarily implies a conscious effort on the part of the minority group to increase its numbers. But paradoxically, Jeffrey and Jeffrey (2005:450) state that in the villages they studied, the higher fertility of Muslims was not a result of conscious religious or political strategies by the Muslims; but Hindu groups were to some extent using fertility for collective purposes! Neither Jeffrey and Jeffrey, nor other proponents of minorityinsecurity hypothesis, provide any evidence to substantiate their thesis. But there are some circumstantial evidences that seriously undermine their argument. The validity of the hypothesis can be checked by examining whether Muslim fertility increases, or declines at a slower rate, during periods when the threat perception is high, or in regions where the incidence of communal riots is more. On both counts, the evidence is to the contrary. As discussed above, the decline of fertility was slower among Muslims than Hindus until the end of 1980s but during the 1990s the pace of reduction was about the same in the two communities. But it was in the 1990s that the Hindunationalist forces gained political ascendancy in India, following the demolition of Babri masjid. If insecurity leads to higher fertility, fertility decline among Muslims should have slowed down during this period. Also, contrary to expectations, Hindu-Muslim fertility differences tend to be larger in states where communal riots are rare, and smaller in states where communal riots are more frequent. For example, the child-woman ratios (CWRs) from the 2001 census show that in Kerala and West Bengal Muslim CWRs are higher by 43-59 per cent, while in Gujarat, Bihar and Uttar Pradesh they are higher only by 6-14 per cent. Thus, taken at its face value, the evidence suggests that the insecurity of minority communities reduces their fertility rather than increases it.

After all, it is not only in India that Muslims have higher fertility. Throughout south-east Asia, Muslim minority groups have higher fertility than the majority community of the respective nations. Even in countries where they are in majority, Muslims are not the forerunners in fertility transition. In Indonesia, Malaysia and

Table 2: Average Annual Growth Rate of Population of Hindus and Muslims during1981-91 and 1991-2001, Major States

Average Annual Growth Rate (Per Cent) Decline in the
Country and State 1981-1991 1991-2001 Growth Rate Difference
Hindu Muslim Hindu Muslim Hindu Muslim (H-M)
India 2.05 2.84 1.82 2.59 0.23 0.25 -0.02
Andhra Pradesh 2.21 2.67 1.35 1.65 0.86 1.02 -0.16
Assam na na 1.39 2.57 na na na
Bihar 2.05 2.59 2.07 3.11 -0.02 -0.53 0.50
Delhi 4.16 6.13 3.65 6.02 0.51 0.12 0.39
Gujarat 1.92 2.15 2.00 2.42 -0.08 -0.26 0.18
Haryana Himachal Pradesh 2.40 1.90 3.78 2.47 2.39 1.57 4.71 2.93 0.01 0.33 -0.93 -0.46 0.94 0.80
Karnataka 1.86 2.43 1.43 2.11 0.43 0.32 0.11
Kerala 1.19 2.27 0.70 1.47 0.48 0.80 -0.31
Madhya Pradesh 2.36 2.72 1.96 2.58 0.40 0.13 0.26
Maharashtra 2.25 2.73 1.96 2.97 0.30 -0.24 0.54
Orissa 1.75 3.14 1.47 2.77 0.28 0.37 -0.09
Punjab* 1.20 3.54 2.53 4.67 -1.33 -1.14 -0.19
(1.86) (1.77) (0.09) (1.05)
Rajasthan Tamil Nadu 2.48 1.41 3.47 1.92 2.46 1.04 3.06 1.28 0.01 0.37 0.41 0.63 -0.39 -0.27
Uttar Pradesh 2.08 3.11 2.16 2.75 -0.09 0.36 -0.45
West Bengal 1.91 3.14 1.33 2.30 0.58 0.84 -0.25

Notes: * The growth rates for the combined population of Hindus and Sikhs are given in parentheses because in the 1991 Census, many Hindus were probably reported as Sikhs. na – not available.

Economic and Political Weekly March 26, 2005

Bangladesh, Chinese and Indian ethnic groups have lower fertility than the Muslims. Thus, for explaining higher fertility of Muslims, a religious basis must be sought. As shown in our paper in the special issue, the percentage of women giving religious reasons for not wanting to use contraception is significantly higher among Muslims than among either Hindus or Christians. The use of sterilisation is particularly low among Muslims. On this issue, it is pertinent to quote Caldwell again: “The pressures within the Muslim population against sterilisation are often coercive. They range from refusal to eat with sterilised, to accept food from them..., to have them as servants, or to allow them to be buried in the Muslim burial ground. The majority of Muslims state the case against sterilisation in terms of religious morality, either that it is forbidden or that Allah has committed himself to provide for children sent to this earth, while contemporary Hindu explanations are completely secular” [Caldwell et al 1982:714]. Theoretically, it is possible for conservative attitudes on fertility control to prevail in communities perceiving external aggression. But in practice, we fail to find evidence for a positive association between communal tension and Muslim fertility, as postulated. It is therefore far more likely that such beliefs are there because many influential clerics genuinely believe that family planning, especially sterilisation, is against Koranic law, and not because of any conscious plot to outnumber Hindus. It is imperative that we make this distinction. In an organised religion with a custom of issuing religious diktats, what religious leaders believe would make all the difference. But, fortunately, religious beliefs tend to lose their hold when more and more members of a community begin to use contraception.

Finally, I wish to point out an error in the procedure used for calculation of net international migration for Hindus and Muslims in the paper by Rajan (2005:440). While the census growth rates he uses are for the decade 1991-2001, the estimates of birth rate he uses are for the latter part of the decade as they were based on CWRs that had children aged 0-6 years in 2001 in the numerator. As the actual birth rates for the decade would have been higher, he overestimates the net international migration into India. Besides, there could be errors in his birth rate estimates because the level and the age pattern of mortality also affect CWRs, and the relationship between the birth rate and CWR is not necessarily linear, as he had assumed.




1 Could be more if the relationship between fertility and years of schooling is concave rather than linear.

2 In a personal discussion, P M Kulkarni has told me that he too had noticed this flaw.

3 There is an alternative explanation, however. It is possible that not all Muslim women disclose the use of contraception to survey interviewers.


Arnold, Fred (1985): ‘Measuring the Effect of Sex Preference on Fertility: The Case of Korea’, Demography, 22(2), pp 280-88.

Bhat, P N Mari and A J Francis Zavier (2005): ‘Role of Religion in Fertility Decline: The Case of India Muslims’, Economic and Political Weekly, 40(5), pp 385-402.

Bhagat, R B and Purujit Praharaj (2005): ‘Hindu-Muslim Fertility Differentials’, Economic and Political Weekly, 40(5), pp 411-18.

Borooah, Vani K and Sriya Iyer (2005): ‘Religion, Literacy and the Female-to-Male Ratio’, Economic and Political Weekly, 40(5), pp 419-27.

Caldwell, John C, P H Reddy and Pat Caldwell (1982): ‘The Causes of Demographic Change in Rural South India: A Micro Approach’, Population and Development Review, 8(4), pp 689-727.

Dharmalingam, A, K Navaneetham and S Philip Morgan (2005): ‘Muslim-Hindu Fertility Differences: Evidence from National Family Health Survey-II’, Economic and Political Weekly, 40(5), pp 429-35.

James, K S and Sajini B Nair (2005): ‘Accelerated Decline in Fertility in India since the 1980s: Trends among Hindus and Muslims’, Economic and Political Weekly, 40(5), 375-83.

Jeffery, Roger and Patricia Jeffery (2005): ‘Saffron Demography, Common Wisdom, Aspirations and Uneven Governmentalities’, Economic and Political Weekly, 40(5), pp 447-53.

Krishnaji, N and K S James (2005): ‘Religion and Fertility: A Comment’, Economic and Political Weekly, 40(5), pp 455-58.

Rajan, S Irudaya (2005): ‘District Level Fertility Estimates for Hindus and Muslims’, Economic and Political Weekly, 40(5), pp 437-46.

Retherford, Robert D and Vinod Mishra (2001): An Evaluation of Recent Estimates of Fertility Trends in India, NFHS Subject Reports No 9, IIPS, Mumbai.


Selections of articles from Economic and Political Weekly

Ideals, Images and Real Lives

Women in Literature and History

Edited by Alice Thorner and Maithreyi Krishnaraj

‘Birds in a Cage’: Changes in Bengali Social Life as Recorded in Autobiographies by Women Sarabashi Ghosh ❖ In Search of the ‘Pure Heathen’: Missionary Women in Nineteenth Century India Geraldine H Forbes ❖ Sarojini Naidu: Romanticism and Resistance Meena Alexander ❖ Women, Emancipation and Equality: Pandita Ramabai’s Contribution to Women’s Cause Meera Kosambi ❖ Outside the Norms: Women Ascetics in Hindu Society CatherineClementin-Ojha ❖ Nationalist Iconography: Image of Women in 19th Century Bengali Literature Tanika Sarkar ❖ Positivism and Nationalism: Womanhood and Crisis in Nationalist Fiction – Bankimchandra’s Anandmath Jasodhara Bagchi ❖ Govardhanram’s Women Sonal Shukla ❖ How Equal? Women in Premchand’s Writings Geetanjali Pandey ❖ Representing Devadasis: ‘Dasigal Mosavalai’ as a Radical Text S Anandhi ❖ The Virangana in North Indian History: Myth and Popular Culture Kathryn Hansen ❖ Construction and Reconstruction of Woman in Gandhi Sujata Patel ❖ Feminine Identity and National Ethos in Indian Calendar Art Patricia Uberoi.

pp xiv + 354 Rs 350

Available from

ORIENT LONGMAN LIMITED Calcutta Chennai Mumbai New Delhi Bangalore Bhubaneshwar Ernakulam Guwahati Hyderabad Jaipur Lucknow Patna

Economic and Political Weekly March 26, 2005

Dear reader,

To continue reading, become a subscriber.

Explore our attractive subscription offers.

Click here


(-) Hide

EPW looks forward to your comments. Please note that comments are moderated as per our comments policy. They may take some time to appear. A comment, if suitable, may be selected for publication in the Letters pages of EPW.

Back to Top