The EAC-PM's Paper on 'Muslim Population' Is a Travesty of Research Practices

Santosh Mehrotra and Bir Singh | 16 May 2024
No image

Fertility rates of the poorest sections of the population of any state are impacted by the quality of educational and health services, and gender differences in regard to health and educational outcome indicators, not religion.

Historically, the role of the Prime Minister’s Economic Advisory Council has been to conduct research and advice the government of India on economic matters of national import. The current government decided to create an EAC-PM in late 2017, almost as an afterthought, over three years after it came to power. Its role, since then, has been to write sundry research papers on miscellaneous topics, regardless of whether they have much to do with economic issues of national import.

This month, a sensational claim is made in a EAC-PM paper – that the Hindu share of India’s population has fallen between 1950 and 2015 by 7.8%, while that of Muslims has increased by 43% over the same period. The national media has gone to town on the subject, feeding as it does the narrative of the party in power and its ideological anchors in a so-called ‘cultural organisation’ – at a critical moment, when the nation goes to the polls, with the government trying with all the means at its disposal, to hold on to power, and win.

The claim is based on the paper authored by Shamika Ravi, et al, which is fraught with many logical and methodological fatal errors. It has used survey data published by the Association of Religion Data Archive (ARDA) consisting of 2,34,520 observations only. There is no mention of the estimated population size of religion-wise communities (The observations provide data on the population and percentages of surveyed people belonging to different religious categories in each country).

An informed reader looks for the coverage, authenticity and accuracy of the data source that has been used to make a claim.

Since census data has all these features rather than survey data, it is more convincing and reliable.

We will therefore proceed to raise a series of questions about the method followed in the paper – which lead to an inevitable conclusion that their findings are questionable.

Why ARDA – and not India’s Census?

First, how have the authors of the EAC-PM reached the conclusion that the Hindu share has shrunk by nearly 8 percentage points (from 84 to 76)?

This change in share of population of Hindus and Muslims during 1950 to 2015 is shown without mentioning their size either at the beginning or the end of the period. We know the size in millions as it is provided by the Census  of India data – but curiously that is not the data source used by the authors. In addition, the paper has not used other vital indicators of India’s demography before making such a claim.

For India, the comprehensive and most accurate data on population measures is collected and published by the Census Office. As per  the Census of 2011, the share of the Hindus and Muslim population is about 79.8% (966 million) and 14.2% (172 million) respectively in the year 2011 (see Figure 1 above).

By contrast, the EAC-PM paper mentions the share of Hindus for 1950 as 84.68% (a full 4.9% higher than the Census’ own record for 1951) and in 2015 as 78.06%. This leads to an estimate of a much larger decline of the Hindu share of the population than is warranted by the government of India’s own data – and which is a politically fraught subject, especially at election time, a fact that the researchers could not possibly have been unaware of.

Perhaps this was done to keep the source the same across countries, since this is a cross-country study. However, despite the EAC-PM advising a government that constantly complains that international sources should not be used for comparisons with India, the authors are undeterred.

Table 1: India’s population by religious groups, 1951-2011 (million)

Census Years    Total Population    Hindus    Muslims    Christians    Sikhs    Buddhist    Jain

1951    361    304    35    8.3    6.8    2.7    1.6

1961    439    367    47    10    7    3.2    2

1971    548    453    61    14    10    3.9    2.6

1981    683    562    80    16    13    4.7    3.2

1991    846    690    107    19    16    6.4    3.3

2001    1029    828    138    24    19    7.9    4.2

2011    1211    966    172    27    20    8.4    4.5

Source: Census of India, various issues, Registrar General of India, Government of India.

It has been claimed by the EAC-PM paper that the majority population share has declined by 7.81% over this period, but India’s Census data shows that the decline of the Hindu population share is actually 4.3 percentage points over 1951-2011 (and not the 7.83% claimed by the EAC-PM paper).  

A related issue is that they are using survey data, which by definition cannot be as accurate as Census data.     

We can understand that the authors used ARDA data for reasons of maintaining comparability across data sources, but by doing that they come to conclusions that unfortunately open them up to multiple and further questions.

The shares of respective populations of religious groups they arrive at are very different from those the Census of India gives us.

Table 2: Percentage share of population by religious groups

Census Years    % Hindus population     % Muslims population    %Christians    %Sikhs    %Buddhist    %Jain

1951    84.1    9.8    2.3    1.79    0.74    0.46

1961    83.5    10.7    2.44    1.79    0.74    0.46

1971    82.7    11.2    2.6    1.89    0.7    0.48

1981    82.3    11.8    2.44    1.92    0.7    0.47

1991    81.5    12.6    2.32    1.94    0.77    0.4

2001    80.5    13.4    2.34    1.87    0.77    0.41

2011    79.8    14.2    2.3    1.72    0.7    0.37

Change    -4.3    4.4    0    -0.07    -0.04    -0.09

% change    -5.11    44.89    0    -3.91    -5.40    -19.56

Source: Census of India, various issues, Registrar General of India, Government of India.

By using survey data, rather than India’s own Census data to get actual, hard numbers, the authors end up committing a third gross error. For the Muslim population they arrive at a share of the population which is way out of line with reality – it exaggerates the decrease in the Hindu share significantly more than that indicated by the Census.

There is an additional problem with this analysis. The authors actually arrive at a 7.81% exaggerated decline in Hindu share of India’s population, because they find that the ARDA survey data shows that the Christian and Sikh population share has also increased at the expense of that of Hindus. This claim ignores an elementary fact: that we have known for some decades that the Sikh and Christian population since 1992 (the first NFHS) have had a much lower fertility rate than Hindus (see Table 2) for the last three decades or more.

How can Sikh and Christian population be pushing down the Hindu share in the population if their TFR has been consistently lower than that of all other communities?

Ignoring demographic, educational and poverty as determinants of fertility rates

There is a third problem with the PMEAC paper. It ignores a core and well known globally known fact about demographic behaviour on the one hand, and educational and poverty levels, on the other. Those communities that have among the lowest levels of educational attainment, especially among women, have the highest fertility rates. Thus, thanks to NFHS data, consistently available at five yearly intervals since 1992-1993 (NFHS 1 to latest NFHS 5, 2019-2021) we have known for decades that the Scheduled Castes, Scheduled Tribes and Other Backward Classes, as well as Muslims, have had the highest fertility rates.

Table 3 reveals the trends in the fertility rates across religion and castes groups. These three communities are among the most educationally disadvantaged among all social groups in India (relative to other religious groups and to higher caste groups), and historically have also suffered from among the highest poverty rates.

Table 3: Total fertility rate by religion/caste, 1993-2021

NFHS year    All    Hindus    Muslims    SC    ST

1992-93    3.39    3.3    4.41    3.92    3.55

1996-99    2.85    2.78    3.59    3.15    3.06

2005-06    2.68    2.59    3.4    2.92    3.12

2015-16    2.18    2.13    2.62    2.26    2.48

2019-21    1.99    1.94    2.36    2.08    2.09

%change (1992-2021)    -41.30    -41.21    -46.49    -46.94    -41.13

Source:NFHS Rounds,1-5            

No serious researcher will ignore the data on literacy levels and fertility rates and the rate of change in these rates to make a generalised assertion about any community.

Table 4 shows that the rate of  illiteracy has been quite high among Muslims and even after seven decades since independence, it is highest at 48.1% (females) and 37.59% (males) among all religious cohorts.

It is much lower for Hindus, particularly males (29.22%). Clearly, there is a responsibility among Muslim societal leaders that every effort should have been made to ensure all Muslim children were going to school (not merely madrassas), especially, after the government of India since the early 1990s tried to universalise elementary education, and enrolments rose among all social groups, including Muslims.

Table 4: Illiteracy by religious groups (%)

Population    Females    Male

All religious communities    44.02    30.24

Muslims    48.1    37.59

Hindus    44.02    29.22

Jains    15.07    12.14

Christians    28.03    23.22

Sikh    36.71    28.68

Buddhists    34.4    22.13

Other    58.62    40.62

Source: Census data, 2011

Table 5: Fertility rates by religious groups (%)

NFHS year    Muslims    Hindus    Christians    Buddhists    Sikhs    Jains    All India

1992-93    4.4    3.3    2.9    2.9    2.4    2.4    3.4

1996-99    3.6    2.8    2.4    2.1    2.3    1.9    2.8

2005-06    3.4    2.6    2.3    2.3    1.9    1.5    2.7

2015-16    2.6    2.1    2    1.7    1.6    1.2    2.2

2019-21    2.4    1.9    1.88    1.39    1.61    1.6    2

Percent decline (1992-2021)    45    42    35    52    33    33    41

Source: NFHS rounds, 1-5 

Table 5 throws light on fertility rates across religious groups as estimated by the NFHS since 1992. No doubt that fertility rate has been highest for Muslims over this period, but it is no higher than it is for SCs or for STs (Table 3) – which is the point we made earlier that the poorest populations of the world have always had the highest population growth rates and higher fertility rates than the better educated and better off. 

But if we look at the percentage decline over 1992 to 2019, we find that the decline in the fertility rate for Muslim women has been the highest (45%). This decline is above the national average (41%).Whereas this decline is comparatively low for Hindu women (42%). In light of this difference in the degree of fertility decline, Muslim population will stabilise earlier than the Hindu population. 

States: How they differ in fertility for different social groups

A final point is notable. Fertility rates of poorest sections of the population of any state, are impacted by the quality of educational services, health services and gender differences in regard to health and educational outcome indicators.

Let us explain with a simple example. A detailed examination of state wise TFRs of various marginalised social groups will make clear why the TFR of Muslim women in both Kashmir and Kerala are lower than the TFR of ‘upper’ caste Hindus of Uttar Pradesh or Bihar. We had demonstrated this fact in the second National Human Development Report for the Planning Commission.

All ideologues who want to make political capital out of the higher than national average TFR of Muslims should ponder this fact. In whichever state you examine, if the quality of public health services and public education is superior, most social outcome indicators of Muslims, as well as SCs and STs, are better than in other states where none of these conditions hold.

States that need to, for the sake of India’s future, bring down their fertility rates for all women – need not waste their time and energy in stirring up political cauldrons for political gain, but rather just ensure that that the public education and health services, especially mother and child health, and women’s education, improves rapidly. Uttar Pradesh, Bihar and Jharkhand should consider these as their top priorities – not whether they become $1 trillion economies. The latter will follow in due course, with inclusive and sustained economic growth.

Santosh Mehrotra is an independent human development economist. 

Bir Singh is Associate Professor of Economics in Delhi University.

This article was originally published on The Wire.
Views in this article are author’s own and do not necessarily reflect CGS policy.