The Demography of South Asia from the 1950s to the 2000sA Summary of Changes and a Statistical Assessment
The countries of South Asia (Afghanistan, Bangladesh, Bhutan, India, the Maldives, Nepal, Pakistan and Sri Lanka) cover less than 4% of the Earth's surface, but their combined population of some 1.6 billion inhabitants in 2007 represents nearly a quarter of the world total. India, the largest country in the region, alone has 1.17 billion inhabitants. This chronicle charts the main demographic trends since the 1950s, which are explained in part by the countries' diverse levels of development. Their demographic transitions also exhibit broad diversity. There is no single transition model specific to the region, just as there is no single transition in India, as the comparison of its states makes clear. Except in Sri Lanka, where the process is complete, the fertility transition is ongoing, and the mortality transition is in general very advanced. The potential for demographic growth remains high in South Asia, and the United Nations expects the region's population to grow by 600 million inhabitants up to 2040. The future course of demographic change has major implications for development, since most of the countries need to reduce poverty and raise educational levels while at the same time coping with rapid urban growth and addressing environmental issues.
Avec quelque 1,6 milliard d'habitants en 2007, l'Asie du Sud (Afghanistan, Bangladesh, Bhoutan, Inde, Maldives, Népal, Pakistan et Sri Lanka) rassemble sur moins de 4 % de la superficie totale de la terre près du quart de la population mondiale et l'Inde, plus grand pays de cette région, compte à elle seule 1,17 milliard d'habitants. Cette chronique présente l'évolution, depuis les années 1950, des principaux indicateurs démographiques, qui s'explique en partie par le niveau de développement des différents pays. Les transitions démographiques apparaissent diversifiées : il n'existe pas un modèle de transition qui serait propre à la région, pas plus qu'il n'existe un modèle indien de transition comme le montrent, pour ce pays, les comparaisons entre États. À l'exclusion du Sri Lanka où elle est achevée, la transition de la fécondité est toujours en cours alors que celle de la mortalité est généralement bien avancée. Le potentiel de croissance démographique de l'Asie du Sud reste très élevé puisque, selon les Nations unies, la population de cette région devrait s'accroître de 600 millions d'habitants d'ici 2040. Les évolutions démographiques futures constituent un enjeu majeur de développement, puisque la plupart des pays doivent réduire la pauvreté, améliorer les niveaux d'instruction, gérer une forte croissance urbaine et mieux protéger l'environnement.
Con cerca de 1 600 millones de habitantes en 2007, Asia del Sur (Afganistán, Bangladesh, Bhután, India, Maldivas, Nepal, Pakistán y Sri Lanka) reúne en menos del 4% de la superficie total de la tierra cerca de la cuarta parte de la población mundial y la India, país más grande de esta región, cuenta, éste sólo, con 1 170 millones de habitantes. Esta crónica presenta la evolución, desde los años 1950, de los principales indicadores demográficos, que se explica en parte por el nivel de desarrollo de los diferentes países. Las transiciones demográficas aparecen diversificadas. En efecto, no existe un modelo de transición específico de la región, ni tampoco existe un modelo indio de transición como lo muestran, para este país, las comparaciones entre Estados. Exceptuando a Sri Lanka en donde se encuentra acabada, la transición de la fecundidad está siempre en curso mientras que la de la mortalidad está generalmente bien avanzada. El potencial de crecimiento demográfico de Asia del Sur sigue siendo muy elevado puesto que, según las Naciones unidas, la población de esta región debería aumentar de 600 millones de habitantes de ahora a 2040. Las evoluciones demográficas futuras constituyen un reto mayor de desarrollo puesto que la mayoría de países deben reducir la pobreza, mejorar los niveles de instrucción, manejar un fuerte crecimiento urbano y proteger mejor el medio ambiente.
South Asia, for the purposes of this chronicle, is composed of eight countries, namely Afghanistan, Bangladesh, Bhutan, India, the Maldives, Nepal, Pakistan and Sri Lanka(1). In 2007 the population in this region totalled nearly 1.6 billion inhabitants (United Nations, 2007a), representing nearly 24% of the world's total population spread over less than 4% of the world's surface(2). From a demographic point of view, South Asia is dominated by India, which alone has a population of 1.17 billion, but there are two other densely populated countries in this region: Pakistan and Bangladesh, with populations of 164 and 159 million respectively in 2007. South Asia also includes countries with small populations such as Bhutan, with only 658,000 inhabitants and the Maldives with just 306,000.
There are sharp contrasts between these countries, be it in terms of demographic growth, population density, mortality and fertility rates, urbanization or literacy. And the situation within India, the largest country in the region, is itself one of major internal contrasts. Its 28 states each have distinct demographic traits – population size, stage of demographic transition, density etc. – as well as different economic, social and religious characteristics.
The demographic challenges confronting South Asia are those of developing countries faced with a major population increase. They are the challenges of educating, housing, caring for and employing a growing – sometimes rapidly growing – population. These countries have to fight against poverty while ensuring that the economic growth they so badly need to improve the lives of their people does not result in serious environmental damage. The populations [End Page 9] of South Asia are very vulnerable both economically (especially the children who start working at an early age) and environmentally. For instance, a portion of the Bangladeshi population lives at sea level with considerable exposure to flood risk, and several countries in the region were affected by the tsunami in the winter of 2004. In certain fragile or exposed areas, demographic growth and the consequent increase in population density exacerbate the vulnerability of the poorest populations.
I. South Asia in the world
With just under one quarter of the world's population in the early 2000s, South Asia's population growth of 2.1% over the past thirty years is far higher than the world average of 1.6% between the years 1975-2005 (Table 1), but below that of sub-Saharan Africa (2.7%) or the Arab world (2.6%). South Asia is less advanced in its demographic transition than the world as a whole: the fertility rate was 3.2 children per woman between 2000-2005 compared with a global average of 2.7, while life expectancy at birth was 3.6 years below the world average. South Asia is quite distinct from East Asia and the Pacific region,
[End Page 10]
Selected socio-demographic characteristics of South Asia compared with other world regions
|Per capita GNI
in USD (PPP)
in 2004 (%)
|TFR: total fertility rate.|
|HDI: Human development index.|
|PPP: Purchasing power parity.|
|Source: UNDP (2006).|
|North Africa and the Middle East||310.5||3.7||67.3||2.6||0.680||5,680||69.9|
|Latin America and the Caribbean||548.3||2.6||72.2||1.9||0.795||7,964||90.2|
|East Asia and the Pacific||1,944.0||1.9||70.8||1.4||0.760||5,872||90.7|
[End Page 11]
where fertility is below replacement level and life expectancy 7 years above that of South Asia. The health situation in South Asia is worse than in other regions of the world, with the exception of sub-Saharan Africa.
South Asia is also the second least developed region in the world after sub-Saharan Africa. In 2004, the human development index (HDI)(3) was 0.60 compared with a world average of 0.74; per capita GDP (USD 3,072 per head in terms of purchasing power parity) was nearly three times below the world average, and adult literacy the lowest in the world, even below that of sub-Saharan Africa.
II. Abundant demographic data in view of development levels
Overall, with the exception of Bhutan and Afghanistan for which demographic information – apart from UN data – is fragmentary, rich sources of demographic data are available for South Asian countries from both censuses and surveys (see Table 2a). The three most densely populated countries, India, Pakistan and Bangladesh, have taken regular censuses since the end of the nineteenth century. India has held one every ten years since 1871. The first post-Independence census was conducted in 1951, while the most recent was in 2001. Pakistan took its first post-Partition census in 1951 and intended to apply the same decennial principle. The second census was indeed held in 1961 but the third was postponed until 1972 because of the war with India. The fourth was held in 1981 but the fifth census was delayed until 1998. After its independence from Pakistan in 1971, Bangladesh held four censuses in 1974, 1981, 1991 and 2001 respectively. Afghanistan held a partial census in 1979, but the one planned for 2004 was postponed and will probably take place in 2008. Nepal has held ten censuses since 1911, the most recent being in 2001. Bhutan has taken a series of censuses but these are really citizenship censuses for checking the nationality of populations in the wake of intense migration. However, the government carried out a full population census in 2005 to obtain reliable socio-demographic data for the country as a whole.
In five countries out of eight, census information is complemented by demographic and health surveys. In the past fifteen years India has carried out three such surveys, called National Family and Health Surveys, and has just published the results of the most recent one held in 2005-2006. Since the early 1990s, Bangladesh has carried out seven demographic and health surveys, five of which were standard, while Sri Lanka carried out three (Table 2b). Two demographic and health surveys were carried out in Pakistan: a first one in 1990-1991 and a second in 2006, but the results of the latest one are not yet [End Page 12] available. The remaining countries in the region (Afghanistan, Bhutan and the Maldives) have never conducted a survey of this type.
Number of population censuses and national demographic surveys carried out since the 1960s in South Asia
|* 1975-1976 World Fertility Survey.|
|** 2002-2003 World Health Survey.|
health surveys (DHS)
Dates of DHS surveys in South Asian countries
|* Maternal and Child Health, Services Provision Assessment Survey.|
|** In-depth study to find out why women wanting to space births or limit family size were not using contraception.|
|*** Surveys featured on the Sri Lanka Department of Census and Statistics website, but not on the Macro international site.|
|Sources: www.measuresdhs.com/countries; www.statistics.gov.lk|
1999-2000 (MCH SPA)*
2001 (maternal mortality)
|Sri Lanka||1987 (standard)
The most densely populated country in the region is also the one for which we have the most demographic data, thanks in part to India's Sample Registration System (SRS) which constitutes a valuable additional source. This system for registering demographic events was set up by the Registrar General between 1964-1965, originally for a few states, and then extended throughout India [End Page 13] between 1969-1970. It is based on a dual system of birth and death records(4). The SRS Bulletins, published twice a year, provide estimates of crude birth and death rates, natural growth rates and infant mortality rates for both urban and rural areas as well as by state.
III. Demographic and socioeconomic heterogeneity
South Asia is a diversified geographic zone, with sharp contrasts between the mountains of Afghanistan or Nepal, for instance, and the lowlands of the Maldives Islands or the Ganges Delta in Bangladesh. Regional geographic diversity is also to be found within India proper: Himachal Pradesh, a Himalayan state, bears no resemblance to the coastal regions of Kerala or Tamil Nadu. This geographic diversity goes hand in hand with demographic, economic and social heteregoneity.
Large demographic disparities
South Asian countries are of very unequal size (Table 3). The population of the Maldives (306,000 inhabitants) bears no comparison with that of India (1,169 billion inhabitants in 2007). Three of the region's eight countries account for 95% of the total population, and nearly three-quarters of the region's inhabitants live in India. South Asian countries also differ in terms of population density, which ranges from more than 1,100 inhabitants per sq.km in Bangladesh, to an average of just 14 inhabitants per sq.km in Bhutan.
The population of these countries is growing at a variable pace. According to UN estimates, the average rate of population growth in Afghanistan was nearly 3.8% for the period 2000-2005, compared with scarcely more than 0.4% for Sri Lanka. The progress of demographic transition is also different. Infant mortality was estimated at 11‰ in Sri Lanka in the early 2000s, but exceeded 160‰ in Afghanistan, whereas fertility, which fell to only 2 children per woman in Sri Lanka is closer to 7 in Afghanistan.
Within India proper (Map 2), there are considerable demographic disparities in terms of population size, rate of population growth and population density, to name just three criteria. For example, in the 2001 census, the most densely populated state in the country, Uttar Pradesh, had a population of 166 million – larger than that of Pakistan or Bangladesh – whereas the least populated state, Sikkim, had population of scarcely more than 540,000. The Union [End Page 14] territory(5) of the Lakshadweep Islands has the smallest population, with just over 60,000 inhabitants in 2001 (Table 4). In the period 1991-2001, the population in India's various states and territories grew at variable rates, with an annual average of 5% in Nagaland, 3.8% in the Territory of Delhi, 1.8% in Punjab (a rate close to the Indian average, which is 1.93%), 1.3% in Andhra Pradesh and 1.1% in Tamil Nadu. Population density also varies sharply between regions. In 2001 it stood at 324 inhabitants per sq.km for India as a whole, but just 13 per sq.km in Arunachal Pradesh, a state that stretches from the Himalayan mountains to the Brahmaputra Valley, and exceeded 900 inhabitants per sq.km in West Bengal, partly because of the city of Calcutta. Population density is particularly high in the territories of Delhi and Chandigarh, (9,294 and 7,903 inhabitants per sq.km respectively) because of intense urbanization. But urbanization is not the sole reason for high population density. In Kerala, a state where only 26% of the population lives in urban areas, density exceeded 800 inhabitants per sq.km in 2001.
Population, surface area and density in South Asian countries in 2007
(inhab. per sq.km)
|* Total surface area minus inland waters.|
|Sources: United Nations (2007a) and Pison (2007).|
Sharp socioeconomic inequalities
An examination of the three significant variables for a country's development considered to play a vital role in a country's demographic transition, namely literacy, GDP and urbanization, reveals sharp contrasts between the countries. For example, while nearly all women over the age of 15 years are literate in the Maldives (96%) and in Sri Lanka (nearly 90%), the literacy rate is below 50% in India, under 40% in Bangladesh, 35% in Pakistan and Nepal, and as low as 13% in Afghanistan (Table 5). Likewise, there is a three-fold difference in per capita income between Nepal (USD 1,490 in terms of purchasing power [End Page 15] parity) and Sri Lanka (USD 4,390). And although the countries in the region are not yet heavily urbanized, considerable discrepancies exist nonetheless. The proportion of city-dwellers in Pakistan, the most urbanized country in the region, is three times higher than in Bhutan (36% versus 12%).
Like the demographic indicators, the socioeconomic indicators vary from one region to another in India. According to data from the 2001 census, with respect to the Indian average of 54%, female literacy rates range from below 34% in Bihar and approximately 88% in Kerala. The wealth of the states varies [End Page 16]
The 28 states and 7 union territories of India by population in 2001
|State or union territory||Population||Population density
(Inhabitants per sq.km)
|* Union territories.|
|Source: Banthia (2001), http://www.censusindia.net|
|Daman & Diu*||158,204||1,411|
|Dadra & Nagar Haveli*||220,490||449|
|Andaman & Nicobar Islands*||356,152||43|
|Jammu and Kashmir||10,143,700||99|
enormously as does the proportion of city-dwellers. In Goa, India's most urbanized territory, the proportion is nearly 50%, compared with the other extreme of 10% in Himachal Pradesh. [End Page 17]
Literacy, standard of living and urbanization in South Asian countries
of women aged 15
in the 2000s (%)
|Per capita GNI
in USD (PPP)
|Sources: UNDP (2006), United Nations (2008).|
Population breakdown by religion is also highly diverse in South Asian countries, as shown in Table 6. In Bangladesh and Pakistan, the vast majority of the population is Muslim. In 2001, 90% of Bangladeshis were Muslim and, in 1998, more than 96% of Pakistanis. The population of Nepal is more than 80% Hindu, whereas Buddhism, the second religion in order of importance, now only accounts for just over 10% of the total population. Hinduism is also the majority religion in India (81%). There are other religions in India but with the exception of Islam (13% of the population), they are very much a minority. The proportion of Christians is just above 2% and that of Sikhs is even lower, while Buddhists account for less than 1% of the total Indian population.
The other countries do not have detailed statistics on their religious composition. Afghanistan is an Islamic Republic and approximately four fifths
Population breakdown by religion in five South Asian countries (%)
|(a) Other religion or non-specified.|
|(b) Of which 3.6% Kiranti.|
|Sources: Census of India, 2001; Census of Pakistan, 1998; Census of Sri Lanka, 2001; Census of Nepal, 2001.|
|Sri Lanka (2001)||7.8||8.5||76.7||7.0||–||–||–|
[End Page 18]
of its inhabitants are Muslim. In Bhutan, Mahayana Buddhism(6) is the state religion. The law in Bhutan guarantees religious freedom, and Hinduism is practiced in certain areas of the country. This freedom is somewhat theoretical, however, with the government preventing the establishment of non-Buddhist missions in the country or the construction of non-Buddhist religious buildings. The Nepalese constitution describes the country as a "Hindu Kingdom" but Hinduism is not a state religion and in 2006 the House of Representatives declared Nepal to be a "secular state". Islam is the state religion in the Maldives, whereas in Sri Lanka more than three-quarters of the population on the island is Buddhist. Here the Muslim, Christian and Hindu communities each account for just 8% of the total population.
Although four-fifths of the population of India is Hindu, the religious composition varies between states, and Hinduism is not the majority religion everywhere. The proportion is over 94% in Orissa, Chhattisgarh and in Himachal Pradesh (where it even exceeds 95%) but Hindus only represent 8% of the population in Nagaland and 4% in Mizoram which both have a majority Christian population. The proportion of Muslims, which was slightly over 13% in the country as a whole in 2001, is under 2% in Punjab, but close to 25% in Kerala and West Bengal, more than 30% in Assam, 67% in Jammu and Kashmir and 95% in the Lakshadweep Islands. There are few Christians in India; nationally the figure stands at 2.3%, but this rises to 19% in Kerala, 27% in Goa, 70% in Meghalaya and 90% in Nagaland. Sikhs, who represent less than 2% of the Indian population, constitute 16% of the population in Chandigarh and 60% in Punjab. In India less than 1% of the population is Buddhist, but in Sikkim Buddhists account for more than 28%.
IV. A history strongly marked by population movements
Frequent migration waves in the past
Major population movements have occurred in the course of South Asia's history, which explains the diversity of the populations in the various countries of the region. Afghanistan, located in the main path of the Central Asian invasions of south or southwest Asia(7), has a Pashtun majority but is also peopled by Tajiks, Hazaras, Uzbeks, as well as by Aimak, Turkmen and Baluch. [End Page 19] The Nepalese are the product of three major migrations(8) from India, Tibet and Central Asia, while Bangladesh was settled by Turks, Arabs, Pashtun and Persians. Sri Lanka was uninhabited until a Prince of Bengal settled there around 540 BC with a suite of 700 persons who became the ancestors of the Singhalese people from whom most contemporary Sri Lankans are descended (CICRED, 1974b).
From ancient times, many great civilizations occupied the vast territory that was India before the partition of West and East Pakistan in 1947(9). Population movements have played a major role in the rise and fall of civilizations. In the beginning of the third millennium BC, the Indus civilization flourished and extended from the Indian Punjab to the Arabian Sea. It derived its strength from its productive agriculture based on the rich alluvial soil and abundant water supplies, which provided numerous crops including cotton. Towns developed and there were numerous exchanges with Western civilizations, especially with Sumer and Egypt. This civilization disappeared around 1500 BC, probably because of the Aryan invasion or natural disasters(10) (ESCAP, 1982). Around 500 BC, the Aryan civilization developed and spread eastwards to the Ganges Valley. Agriculture also flourished and the population grew, partly as a result of new waves of Aryan migration from Central Asia. The Aryans pushed the Dravidians, who originally inhabited the Indus Valley, further south and the Dravidian civilization subsequently flourished in southern India. For a long time southern India was divided into numerous kingdoms engaged in incessant war against each other.
In the north of India, the Maurya Empire extended from Afghanistan to Bengal via the southern Himalayas, and reached its apogee under Ashoka (273-232 BC). Then, from the early fourth century AD, the Gupta Empire extended from the Indus to the Brahmaputra Rivers to include a large part of the Indian peninsula. This region subsequently enjoyed 200 years of peace and political stability up to the invasion of the Huns at the end of the fifth century. The Mogul Empire followed from the sixteenth to seventeenth century, after which the Portuguese, Dutch, British and French traders arrived. The East India Company(11) established during the Mogul period greatly advanced Britain's domination of India. [End Page 20]
The British influence
All the countries of South Asia have been subjected to strong British influence. In Afghanistan, the British fought the Russians for control of the country from the end of the 1830s. Between 1838 and 1842, the first Afghan war ended with the defeat of the British army, but in the second Afghan war of 1878-1880 the British army conquered Kabul and freed Kandahar. In 1919, a third Afghan war resulted in that country's independence. After a war that was intended to control the expansionist aims of the Gurkhas, Nepal came under British rule in 1815 and remained so until 1923, when Great Britain recognized Nepal's independence. The annexation of Assam by the British in the 1820s stirred up tensions on the Bhutanese border and conflicts ensued until a treaty was signed in 1910 that made Bhutan a British protectorate. From the mid twentieth century, Bhutan was largely under Indian dominance while Sri Lanka was under British rule from 1815 to 1948. It was the British who introduced tea plantations to the island that was then called Ceylon. The Maldives were a British protectorate from 1887 to 1965.
India was almost entirely administered by the British from the mid nineteenth century until independence in 1947(12). The British unified the country, in particular by constructing a railway network. They carried out irrigation work, set up communications networks, developed the mining industry, cultivated jute and tea, etc. To get a clearer picture of the population on its territories, the British government conducted a first census, known as the 1872 census, but which in fact stretched over the period 1867-1872. The first detailed "instant" census was conducted in 1881. From that time until the present day, colonial India and later independent India has taken a census every decade, the most recent in 2001.
Population diversity and political tensions
The political history of this region, like the diversity of its populations in terms of origin or religion, has given rise to numerous conflicts, some of which continue to this day.
The independence of India and the partition of East and West Pakistan in 1947, resulted in major population movements, with Hindus migrating from Pakistan to India and Muslims settling in Pakistan(13). In 1971, East Pakistan fought a "liberation war" with West Pakistan, accused of political and economic dominance, resulting in East Pakistan's secession and the formation of Bangladesh in December of the same year. The problem of Kashmir remains a source of friction between India and Pakistan, who have clashed three times on that issue. Any changing or crossing of borders between Indian and Pakistani [End Page 21] Kashmir always leads to confrontations, particularly since the Muslims on the Pakistani side tend to be increasingly radical. Afghanistan has also been at the heart of a series of conflicts over the past thirty years, with the Soviet invasion of 1979, the takeover by the Taliban in 1996, the toppling of their regime in the 2001 war, and incessant guerrilla warfare ever since. Sri Lanka has seen violent confrontations between the majority Singhalese population and the Tamils. The Maldives became a republic in 1968, but endured a coup 20 years later during which the Maldivians obtained help from the Tamils, themselves in conflict with the Sri Lankan authorities. However, with Indian help, the Maldives government resisted the coup. Bhutan suffered from ethnic violence between 1991-1993 and in India violence regularly erupts between various communities. The Indian government had to contain separatist Sikhs in the 1980s and considerable tensions have existed between fundamentalist Hindus and Muslims for nearly 20 years. Bangladesh is also subject to Islamist violence.
V. Population and growth since 1950
Sustained population growth
In the early 1950s, the demographic growth rate in the South Asian countries (Appendix Table A.3) was twice as high in some countries as in others, ranging from 1.54% in Afghanistan to 3.09% in Bhutan. Several countries were at the higher end of the scale – Sri Lanka, with annual population growth of 2.8%, was not far behind Bhutan, and Pakistan, though growing more slowly, was still above 2%. During the following decades, all the countries except Sri Lanka witnessed an acceleration of their demographic growth. The annual growth rate in India rose from 1.73% in 1950-1955 to 2.26% thirty years later, in 1980 1985. During the same period, that of Bangladesh increased from 1.98% to 2.47% and that of Pakistan from 2.15% to 3.63%. Between 1980-1985 and 2000-2005, all the countries, with the exception of Afghanistan, recorded a slowdown in population growth (Figure 1). In Afghanistan, over the period 1980-1985, the natural growth rate was largely positive, but the war with the Soviet Union prompted massive migration of refugees to Pakistan and Iran that led to negative total population growth(14).
Sri Lanka has proved to be an exception, with a steady decrease in the population growth rate since the early 1950s: from 2.8% in 1950-1955, it dropped to 2.1% in 1970-1975, 1.4% in 1980-1985, 1.1% in 1990-1995 and finally 0.4% in 2000-2005, i.e. a decline to virtually zero population growth in the space of fifty years (Appendix Table A.3).
India, which set up its first family planning programme in the early 1950s has seen its population growth rate slow down over the past twenty years, [End Page 22] though in 1990-1995 it was still above 2% per annum. Although the Indian authorities have been trying to stabilize the population growth rate for many years now, the figure was still 1.62% per annum in 2000-2005. During the same period, the population growth rates in Pakistan and Bangladesh remained above 1.8%, which means that if there is no change, the population of these countries will double in less than forty years. As a result of these high demographic growth rates over a period of a half a century, the population of South Asia tripled between 1950 and 2005, climbing from 478 million to 1,580 million inhabitants.
South Asia in 2040: a forecast increase of 600 million inhabitants
According to the medium variant projections of the United Nations (2006 revision), the population in the region will continue to grow during the next few decades, reaching a total 2,200 million inhabitants in 2040, i.e. 600 million inhabitants more than today. Its share in the total world population will increase slightly, from 23.3% in 2005 to 25.3% in 2040. According to these mean variant projections, in thirty years time, India will have a population of 1,600 million inhabitants, Pakistan 268 million, Bangladesh 239 million, Afghanistan 66 million, Nepal 47 million, Sri Lanka about 20 million, Bhutan 900,000 and the Maldives 476,000. By then, India will be more populated than China, with its estimated 1,450 million inhabitants. [End Page 23]
Projections drawn up in India on the basis of the 2001 census and revised in 2006 lead to a slightly lower population in 2025 than the figure put forward by the United Nations (1,389 million against 1,447 million), but the difference – nearly 60 million – needs to be put into perspective, as it only corresponds to slightly more than two years of births in India. In 2026, the final year of these projections, it is forecast that India will have about 1,400 million inhabitants (Table 7). The most populated state, Uttar Pradesh, will have a population of 249 million inhabitants, equivalent to the entire population of India in 1911. Three other States will have more than 100 million inhabitants: Maharashtra (133 million), Bihar (114 million) and Western Bengal (101 million). Andra Pradesh, Madhya Pradesh(15) and Rajasthan will have 94 million, 88 million and 82 million inhabitants in 2026.
Population of Indian States of more than 20 million inhabitants in 2001 (in thousands) projected to 1 March 2026
|State||Projected population in 2026 (thousands)|
|Source: Office of Registrar General & Census Commissioner (2006).http://www.jsk.gov.in/projection_report_december2006.pdf|
[End Page 24]
VI. Diversified demographic transitions
A comparison of demographic transition "models" over half a century (Figure 2) does not indicate that there is a type of transition specific to the region; neither does it show any highly contrasted transitions between the different countries. The only models that really stand out are those of Afghanistan and Sri Lanka, the first because of its high birth rate and the second because of its sharp fertility decline.
According to the United Nations data – 2006 revision (2007a) – the mortality rates vary considerably between the two most extreme countries (Appendix Table A.3). In the first half of the 1950s, the overall mortality rate was nearly 37‰ in Afghanistan, against a little more than 11‰ in Sri Lanka. In the other countries of the region, it varied from 23‰ to 28‰ at the beginning of the 1950s. A general drop in mortality occurred and the crude mortality rates for all countries in 2000 to 2005 were between 6.5‰ and 8.7‰, with the exception of Afghanistan, with an estimated rate of 21.6‰ for the same period.
Afghanistan and Sri Lanka also represent extreme situations in terms of birth rates. There has been little change in Afghanistan over the last fifty years, with a gross estimated rate of 52.1‰ in 1950-1955 and 49.7‰ in 2000-2005. On the other hand, during the same period, the birth rate continued to fall in Sri Lanka and was more than halved between 1950-1955 and 2000-2005, dropping from 39.9‰ to 16.3‰. In the other countries in the region, the birth rate, still very high at the beginning of the 1950s (between 43‰ and 49‰), also dropped by half in fifty years, to between 22‰ and 30‰ in 2000-2005, levels that remain well above that of Sri Lanka. Only this last country (Figure 1) is in the process of completing its demographic transition.
Although India registered a downward trend in its birth rate over half a century, it is still far short of the target fixed by the Indian government in each five-year plan, which means that birth rate and fertility targets are frequently revised (Rajan and Véron, 2006). The fact that India has a young population maintains birth rates at high levels: this age structure effect (the number of women of child-bearing age is high) explains why the overall drop in fertility only partially leads to a decrease in the birth rate. But observed fertility is also higher than targeted fertility, due to limited access to contraception, particularly in the poorest areas where the population is less educated.
The demographic transition in the South Asian countries over the past half century can also be analysed in terms of the demographic growth rates at the beginning and end of the period. In Afghanistan, for example, the annual population growth rate increased from 1.54% in 1950-1955 to 2.81% in 2000 2005, due to the fairly regular decrease in the death rate, while the birth rate remained practically constant (Figure 2A). On the other hand, in Bangladesh (Figure 2B), the growth rate remained practically the same (1.99% at the beginning and 1.95% at the end of the period). The downward trend in the [End Page 25]
[End Page 26]
death and birth rates was of the same order, although the drop in mortality took place earlier. The decrease in the birth rate was subsequently more rapid, however. In Bhutan (Figure 2C), the growth rate in 1950-1955, the same as that of Bangladesh (1.99%) at the beginning of the period, fell sharply to 1.46% in 2000-2005. In Bhutan, the birth rate dropped more quickly than the death rate. In India (Figure 2D), the growth rate is less than 0.1 point lower than it was fifty years ago. During the last half-century, in fact, the population growth of India increased substantially, reaching 2.3% in 1980-1985, but then slowed down considerably (decrease of more than 0.6 points). In the Maldives (Figure 2E), the population growth rate dropped 0.2 points in the fifty year period: the mortality rate decreased regularly while the birth rate remained high for a long period before dropping drastically over the last twenty years. In Nepal (Figure 2F), the population growth rate increased sharply, from 1.58% to 2.15%, with the death rate dropping more quickly than the birth rate, even if the latter has picked up over the last ten years. In Pakistan (Figure 2G), the population growth rate dropped slightly, from 2.14% at the beginning of the 1950s to 1.98% at the start of 2000. The transition had already begun in Sri Lanka (Figure 2H) in the 1950s, which explains the rapid population growth at that time (2.85%). The death rate was already on the decline, while the birth rate was still high. Today, the transition has been completed and the population growth rate in Sri Lanka is below 1%.
VII. A continuing tradition of early and intense nuptiality
Female nuptiality, which was particularly early in India at the end of the nineteenth century, with a mean age at first marriage of only 13 in 1886-1891, was still early in the 1950s, despite a marked increase in age (16 in the period from 1951-1961) (ESCAP, 1982). Men married much later, at a mean age of 20 in 1886-1891 and 22.3 in 1951-1961. In Bangladesh and Pakistan, female nuptiality was still very early at the beginning of the 1950s, with a mean age at first marriage of 14.4 years, differing considerably from that of men, who married at a mean age of 22.4 years (ESCAP, 1981). Data concerning the 1960s in Karachi indicate much later female nuptiality in urban areas (average age of 19 in 1962) and little difference between social groups (CICRED, 1974a). In the decade from 1961 to 1971, the mean age at first marriage for women was estimated at 17 in India(16). In Sri Lanka, it was already much higher in the 1960s than in the other countries, with a mean age of 23 in 1961 (CICRED, 1974b).
The countries where recent nuptiality data is available (Bangladesh, India and Nepal) are characterized by a continuing tradition of very early and [End Page 27]
Percentage of married persons by age, and age at first marriage in Bangladesh, 2004
|Percentage of women married for
the first time by exact age (years)
|Source: DHS 2004.|
Percentage of men married for
the first time by exact age (years)
practically universal female nuptiality (Tables 8a, 8b and 8c). According to the 2004 Bangladesh demographic and health survey, the median age of women at first marriage is below 15 for the 20-49 age group. Nuptiality only appears to be slightly later in recent cohorts (women aged 25-29 at the time of the survey) – 15 years of age against 14 – even if the data are not strictly comparable because first marriages have not necessarily all taken place in the most recent cohorts (if later marriages are numerous, the median age at marriage will be higher). In India, in 2005-2006, the median age at first marriage for women aged 20-50 at the time of the survey is 17.2, i.e. two and a half years more than in Bangladesh. Once again, the median age is higher for the most recent cohorts: for women aged 20-24, it is nearly two years higher than that calculated for [End Page 28]
Percentage of persons married by age, and age at first marriage in India, 2005-2006
|Percentage of women married for
the first time by exact age (years)
|Source: NFHS, 2005-2006.|
|Percentage of men married for
the first time by exact age (years)
women aged 45-49. In Nepal, the median age at first marriage was also 17 in 2006. In Sri Lanka, on the other hand, the nuptiality model is radically different, and women marry at a much later mean age – 22.9 years – in 2003. In fact, this figure had already been reached at the beginning of the 1960s, and since then, the mean age at first marriage for women has fluctuated around 23 years (CICRED, 1974b).
Another way of appreciating the prevalence of early marriage is to consider the proportion of women already married at a specific age: 15, 18, 20, etc. (Tables 8a, 8b and 8c). In Bangladesh, half of all women aged 25-29 in 2004 were already married at 15, against one quarter of Indian women and 13% of Nepalese women in the same age group. These surveys conducted at the [End Page 29] beginning of the year 2000 show that 80% of women aged 25-29 were already married at 18 in Bangladesh, 60% in Nepal and 55% in India. The general trend is towards female marriage at a later age, with the percentage of women married at 15, 18, 20, etc. decreasing from one cohort to the next. For men, the decrease is not as generalized, but a certain increase in the age of male marriage has been observed in India and Nepal.
Percentage of persons married by age, and age at first marriage in Nepal, 2006
|Percentage of women married for
the first time by exact age (years)
|Source: DHS, 2006.|
|Age group (years)||Men|
|Percentage of men married for
the first time by exact age
Traditionally, men marry later than women, In Bangladesh in 2004, 79% of women aged 20-24 were already married at 20 versus only 15% of men of the same age; 30% of men aged 25-29 were married at age 22 while the percentage of women was above 90%. The 2006 survey in Nepal shows that the difference in marriage behaviour between men and women is less marked: 87% of women aged 25-29 and 56% of men were already married at age 22. The nuptiality [End Page 30] model in India is similar to that in Bangladesh, with men marrying much later than women, although the number of men already married at age 20 is higher in India (26%) than in Bangladesh (15%). Regional data on age at first marriage show a broad diversity in nuptiality behaviour in India (NFHS, 2005-2006). While across India, 46% of women aged 18-29 were already married at age 18, the figure was much lower in Goa (12%) and as high as 64% in Bihar. The differences are also large for men: 27% of Indian men aged 21-29 were already married at 21, with a range extending from only 2% in Kerala to more than 49% in Rajasthan.
In Bangladesh, India and Nepal, female celibacy remains extremely rare: the highest proportion of women aged 45-49 who have never married is observed in Nepal where it stands at just 1.2%. Celibacy also remains exceptional among men: the proportion of single men aged 45-49 is in the same order of magnitude as that of women. These figures apply to the older generations: an increase in the age of marriage may be accompanied by an increase in permanent celibacy among the most recent cohorts.
VIII. An ongoing fertility transition
The mean number of children per woman varies by a factor of two
Fertility in all the countries of South Asia was still high in the early 1950s, ranging from 5.7 children per woman in Sri Lanka to 7.7 in Afghanistan (Appendix Table A.5). At that time, fertility rates were comparable in India and Nepal (5.9 and 6.1 children per woman), substantially higher in Pakistan, Bangladesh and Bhutan (6.6-6.7 children per woman), and even higher in the Maldives (7 children per woman) though still below the level for Afghanistan (7.7).
Between 1950-1955 and 2000-2005, fertility declined sharply in all countries of the region, except in Afghanistan where, according to United Nations estimates, it continued to fluctuate between 7.5 and 8 children per woman. Over this fifty-year period, the mean number of children per woman fell by around 40% in Nepal and Pakistan, by 47% in India, by over 50% in Bangladesh and Bhutan, by 60% in the Maldives and by 65% in Sri Lanka. Women in Sri Lanka had slightly over 2 children on average in 2000-2005 and the fertility transition there is complete, whereas in Afghanistan it has yet to start. In the other countries the fertility transition is under way, though the process is more or less advanced depending on the country: today, the mean number of children per woman is 2.8 in the Maldives and 2.9 in Bhutan, 3.1 in India and 3.2 in Bangladesh, 3.7 in Nepal and 4.0 in Pakistan. Afghanistan is very much the exception (Figure 3).
In most of the countries, women still begin their reproductive life very early, a fact related to early marriage and the notable absence of out-of-wedlock [End Page 31] births (Table 9). In 2004, median age at first childbirth in Bangladesh was below 18, by which age nearly 33% of adolescents have already started childbearing. In India, on the basis of the NFHS-3 survey (2005-2006), median age at first childbirth is 20. The proportion of women aged 15-19 who have ever given birth rises rapidly with age: 1.3% at 15 years, 4.1% at 16, 8.6% at 17, 17.9% at 18 and 29.7% at 19(17). In Sri Lanka, however, childbearing begins much later and this is not a recent phenomenon: in 1987, mean age at first childbirth was already 24 years, and in 2000 it was 23 years.
Median age at first birth (women aged 20-49) and proportion of adolescents (women aged 15-19) who have started childbearing
|Median age at first birth (in years)||Proportion of
childbearing (in %)
|* The most recent DHS survey for Pakistan dates from 2006 but the results are not yet available.|
|** Mean age at first birth.|
|Sources: DHS and NFHS-3 surveys.|
|Sri Lanka (2000)||23.2**||–|
The patterns of age-specific fertility in 2000-2005 that emerge from the United Nations data reveal the progress of the fertility transition in each country of the region (Table 10 and Figure 4). The transition has not yet begun in [End Page 32] Afghanistan, is complete in Sri Lanka, and is under way in the other countries of South Asia. The trend in age-specific fertility follows a similar course in the Maldives, India, Bangladesh, Bhutan and Nepal, although the rates at each age may differ. Pakistan differs from the other countries in having a substantially higher level of fertility from the 25-29 age group onwards, reflecting lower levels of contraceptive practice.
Age-specific fertility rates in 2000-2005 in the countries of South Asia (‰)
|Source: United Nations, 2007a.|
Between a country at the end of the demographic transition, like Sri Lanka, and one still engaged in it, like India, the trends in age-specific fertility present [End Page 33] sharp contrasts (Tables 11a and 11b; Figures 5a and 5b). In Sri Lanka, a fertility decline at all ages between 1963 and 2000 was explained by a high level of contraceptive practice (70% contraceptive prevalence in 2000), while relatively low fertility rates at younger ages are due to late marriage: the fertility rate is below 30 per thousand at ages 15-19 and slightly above 80 per thousand at ages 20-24. The main feature of the fertility trend in India is a decline after age 25. The age-specific fertility rate still stands at 90 per thousand at ages 15-19 in 2005-2006 and exceeds 200 per thousand at ages 20-24. The age-specific fertility curve for India is characteristic of a population in which contraceptive practice is well established but where later entry into childbearing, resulting from increased age at marriage and use of contraception in the early years of union, is not yet widespread. We will return to this point below. The fertility rate in Bangladesh is even higher than in India at ages 15-19 (135 per thousand in 2004) but slightly lower at ages 20-24 (192 versus 209 per thousand) and similar at ages 25-29. In Nepal, the fertility rate in 2006 is lower than in Bangladesh at ages 15-19 (close to 100 per thousand) but substantially higher, however, at ages 20-24 (234 per thousand). At ages 25-29, fertility is ten points higher than in India and Bangladesh.
Age-specific fertility rates in Sri Lanka between 1963 and 2000 (‰)
|Sources: World Fertility Survey for 1975; DHS surveys for 1987, 1993 and 2000.|
Age-specific fertility rates in India between 1970-1972 and 2005-2006 (‰)
|Sources: SRS for 1970-1972 and 1980-1982; NFHS for 1992-1993, 1998-1999 and 2005-2006.|
[End Page 34]
The surveys conducted since 2000 reveal fertility differences that vary by place of residence, but most markedly by women's educational level (Table 12). While the urban-rural fertility differences are non-negligible: 0.7 children per woman in Bangladesh in 2004, 0.9 in India in 2005-2006 and 1.2 in Nepal in [End Page 35] 2006, the differences by women's educational level are very wide everywhere. In Bangladesh, women with no education have on average 3.6 children versus only 2.2 for the most educated women. In India, women with no education have identical fertility to comparable women in Bangladesh whereas highly educated women – with 12 years or more of education – currently have only 1.8 children. In Nepal in 2006, there is a difference of two children between women of the lowest and highest educational levels. In this region as in the rest of the world, education is a key variable in the fertility transition.
Total fertility rate by women's place of residence and educational level
|At least 10 years
|* less than 5 years of education, ** 5-7 years of education, ***8-9 years of education, ****10-11 years of education, *****12 years or more of education.|
|Sources: DHS and NFHS surveys.|
The changes in net reproduction rates(18) summarize the changes in mortality and fertility. With a net reproduction rate of 0.96 daughters per woman in 2000-2005 (Table A.5), it appears that the population of Sri Lanka is not quite replacing itself. Over the same period, the net reproduction rate for Afghanistan remained high, at around 2.3. For the other countries, the net reproduction rate is between 1.2 (Maldives and Bhutan) and 1.7 (Pakistan). [End Page 36]
Fertility is linked to social and health development
The inverse relationship between fertility and development is generally confirmed by a country-by-country comparison of total fertility rates for the period 2000-2005 with the human development index for 2004. The countries where fertility is highest tend to be those for which the development indicators are lowest (Figure 6). For example, fertility in Nepal and Pakistan is much higher than in Sri Lanka, while the human development index is much lower. However, this relationship is not very strong: the development indicators for Bhutan and Pakistan are virtually identical whereas the fertility difference between the two countries is of one child per woman; fertility in Bhutan is practically the same as in the Maldives (2.91 and 2.81 children per woman in 2000-2005, respectively), while the human development index shows a substantial difference (0.54 in Bhutan, 0.74 in the Maldives). The coefficient of determination R2 for this relationship is slightly above 0.6.
The correlation – positive this time – between fertility and infant mortality is also observed over the period 2000-2005. In Sri Lanka, low fertility is combined with low infant mortality, while in Afghanistan fertility and infant mortality are both high (Figure 7). The coefficient of determination is close to 0.98, indicating a particularly strong correlation. This positive correlation between fertility and infant mortality is observed in all world regions (Tabutin and Schoumaker, 2004; Tabutin and Schoumaker, 2005; Guzman et al., 2006).
[End Page 37]
Fertility decline is related to economic and social development but can also result, at least partially, from the implementation of population policies. The case of India is interesting in this respect. In the early 1950s it was the first major country to initiate a family planning programme, and then to attempt a veritable population policy (Srinivasan, 1995). The pace of population growth accelerated in subsequent decades, and the targets, notably for the birth rate, proved over-ambitious and had to be revised several times (Rajan and Véron, 2006). This does not mean that the family planning programmes had no effect at all, since at this time India's population was getting younger, with the consequence that the birth rate decreased less than fertility. However, without "modernization"(19), the fertility transition would certainly have been much slower (Visaria, 1995; Visaria and Visaria, 1995).
Wide disparities in fertility still exist in India
At the regional level, the history of the fertility transition in India breaks down into three phases: an early fertility decline confined to Kerala state in the 1960s; emergence of a division between low fertility in the south and high fertility in the north in the 1970s and 1980s; a generalized decline in the subsequent period, though with large differences persisting in 2005. [End Page 38]
Fertility changes in Kerala and in India (number of children per woman)
|Sources: Mari Bhat and Rajan (1997), SRS, NFHS, NFHS-2 and NFHS-3.|
In the 1950s, fertility in India as a whole still stood at 6.3 children per woman(20), but it was already slightly lower – 5.6 children – in the southern Indian state of Kerala (Table 13). Kerala is the state where fertility declined earliest. The total fertility rate fell there to 5 in the period 1961-1971, while for India as a whole it held steady at 6 children per woman. The gap widened in the following years. According to the NFHS-2, fertility in Kerala was below 2 children per woman in the period 1996-1998. Indeed, for 1994 the Sample Registration System gave a value for fertility half that of the national value: 1.7 versus 3.5. The most recent survey (NFHS-3) finds a difference of 0.75 children per woman between Kerala and India as a whole. Many studies have attempted to explain the reasons for the singular course of fertility change in Kerala (Zachariah and Patel, 1984; Mari Bhat and Rajan, 1997). The high educational level of women and the political specificity of Kerala – a state that became communist in 1957 with the aim of reducing socioeconomic inequalities and improving the health of the population – have often been invoked to account for this earlier demographic transition. In reality, change was of an extremely general nature (Rajan and Véron, 2006). Improvements were indeed made in the education of women, but both age at marriage(21) and contraceptive prevalence also increased considerably. Programmes were also set up to promote maternal and child health, and in 1985 the government launched a programme of universal vaccination, one result of which was to make the population more familiar with the health services. This programme seems to have been as important as specific family planning initiatives in making couples give more [End Page 39] thought to the "quality" of children, known to be a factor in fertility reduction (Zachariah et al., 1994). The state of Kerala did not have a particularly advanced level of economic development, nor was it highly urbanized. It offers an example of a transition linked not to economic development or urbanization but to changes in the social sphere, and in this sense it did not follow the standard pattern of transition. Thereafter, Kerala's demographic singularity gave way to a north-south division of India.
Total fertility rate in the northern and southern states of India in the 1990s (number of children per woman)
|Region and state||1980-1982||1990-1992||1996-1998|
|Sources: Visaria (2004), NFHS and NFHS-2.|
The fertility transition in India during the 1990s was characterized by large fertility differences between the country's northern and southern states(22) (Nair and Véron, 2002). At that time, fertility in the southern states – Kerala, Tamil Nadu, Karnataka and Andhra Pradesh – was 2 children per woman, whereas women in the northern states – Uttar Pradesh, Bihar, Madhya Pradesh and Rajasthan – were still producing 3-4 children on average (Table 14). No single factor can account for these contrasts in fertility (Véron, 2000). Levels of infant mortality and fertility are certainly both lower in the southern states, but the changing relationship over time between these two indicators varies from state to state, as the covariation in these indicators over the period 1981-1994 shows. On Figure 8, each dot represents a combination of fertility and infant mortality at a given point in time, and each broken line plots the covariation in fertility and infant mortality in one state between 1981 and 1994. Over this period, the combined fertility and infant mortality values are consistently lower in the four southern states than in the northern states, and [End Page 40]
the two distinct groups are identified clearly in Figure 8. Thus the broken line furthest to the left on Figure 8 represents this covariation of fertility and infant mortality in Kerala between 1981 and 1994. It is seen that in some years fertility has fallen substantially without any clear reduction in infant mortality. Generally speaking, at state level, no close relationship between fertility and infant mortality is observed. Variation in one of these indicators may be accompanied by no clear change, or even by an increase in the other, thus producing a somewhat erratic trajectory over time. The other variables usually considered have equally limited explanatory power. Though the states in the north of India are less urbanized than those in the south, and fertility is invariably lower in urban than in rural areas, the degree of urbanization does not explain the fertility level in a state: Bihar is less urbanized than Uttar Pradesh, yet fertility was lower there in the early 1990s, and Kerala, though less urbanized than Tamil Nadu, also had lower fertility (Table 15). The scale of fertility differences between urban and rural areas varies widely within states. Rural fertility exceeds urban fertility by 8% in Tamil Nadu while the difference is 45% in Uttar Pradesh. A relationship also exists between female level of education and fertility: more educated women have fewer children. On its own, however, the educational level of mothers does not explain the fertility differences: in 1991-1992, fertility among illiterate women ranged, for a national average of 4 children, from 2.31 in Kerala to 5.36 in Uttar Pradesh. Fertility also rises with the proportion of people living below the poverty line. However, comparable proportions living below the poverty line (43% in Madhya [End Page 41] Pradesh and 45% in Tamil Nadu) can be associated with very different fertility levels (3.9 and 2.5 children per woman, respectively). Equally, the differences in fertility between states are not explained by the religious composition of the population, even though at the national level the number of children per woman does vary by religion (with a difference of one additional child for Muslims relative to Hindus in 1990-1992). In Tamil Nadu, Muslim and Hindu women produce the same number of children, whereas the fertility of Hindus in 1990-1992 ranged from 1.7 children per woman in Kerala to 4.7 in Uttar Pradesh(23).
Urbanization and fertility in India in the early 1990s
|Region and state||Percentage urban||Total fertility rate (number
of children per woman)
|Ratio of rural
|Urban areas||Rural areas|
|* Ratio of mean number of children per woman in rural areas to mean number of children per woman in urban areas (as a percentage).|
|Sources: census of 1991; NFHS 1992-1993.|
The most recent data for India, taken from NFHS-3, show persistently strong spatial disparities in fertility, ranging from 1.79 children per woman in Goa and Andhra Pradesh to 4 in Bihar (Table 16). Replacing the north-south division of India by a finer geographical subdivision – that splits the country into north, centre, east, north-east, west and south – does not cause any homogeneous regions to appear although the previous demarcation lines disappear. Thus, under the partitioning for NFHS-3, what is termed the north region contains Himachal Pradesh where fertility was 1.94 children per woman and Rajasthan where it was 3.21 children per woman. In the north-east region, fertility ranges from 2.02 in Sikkim to 3.80 in Meghalaya. The southern states, however, remain relatively homogeneous in terms of fertility, which is everywhere at sub-replacement level. [End Page 42]
Total fertility rate (number of children per woman) in the states of India in 1990-1992, 1996-1998 and 2005-2006
|Source: NFHS, NFHS-2 and NFHS-3.|
|Jammu and Kashmir||–||2.71||2.38|
The stage reached in the fertility transition in each state depends on the level of development, but no single dominant factor can be identified. Thus while the poorest states are indeed those where fertility remains highest, the correlation between the mean number of children per woman and the proportion of the population living below the poverty line is less than 0.40, and the same proportion of people living below the poverty line can be associated with sharply contrasted fertility levels. In the mid 2000s, fertility in India was on average [End Page 43] 2.68 children per woman and 21.8% of the population lived below the poverty line, yet for similar fertility levels, as in Gujarat and Orissa (2.42 and 2.37 children per woman, respectively) the associated proportions in poverty diverged widely: 12.5% of the population in the former and 39.9% in the latter.
The difficulty of identifying simple determinants of fertility at the aggregate level of states was clearly illustrated several years ago by a multivariate analysis of the relationship between fertility and women's education, which also considered the indirect effects of female education via infant mortality and age at marriage (Sharma and Retherford, 1990). The model's explanatory power was not improved by introducing the level of urbanization. When the residuals were mapped, the deviations from the value explained by the model exhibited geographical contiguity by size, suggesting the influence of regional or local factors (religion, customs, and such like). The complexity of the processes at work was highlighted in a geographical analysis of the fertility transition in southern India by C. Z. Guilmoto and I. S. Rajan (2005), which showed that the explanation needs to include family planning programmes, socioeconomic transformations, and changing reproductive attitudes, and take into account local contexts and the changes affecting them. These authors also viewed the reduction in fertility as the consequence of a social change that diffused gradually between neighbouring populations (Guilmoto and Rajan, 1998 and 2001).
IX. Diversity in attitudes and behaviour towards contraception
Varying levels of contraceptive practice
The data on contraception in the countries of South Asia, though variable in both quantity and quality (United Nations, 2005), are adequate to show the diversity of attitudes and behaviour towards contraception (Figure 9). While contraception remains practically non-existent in Afghanistan, where fewer than 5% of women aged 15-49 in unions use any method, contraceptive prevalence in Sri Lanka already stood at the high level of 70% in 2000. The situation in the other countries is also varied: the proportion of women(24) using some form of contraception is 18.8% in Bhutan, 27.6% in Pakistan, 39.3% in Nepal, 48.2% in India and 58.1% in Bangladesh.
The use of modern contraception also varies considerably between countries. In Bhutan, barely 19% of women in a union and of childbearing age use any means of contraception(25), but modern methods are used exclusively: 3% are protected by female sterilization, 8% by male sterilization, 2% by the pill, 4% by injection or implant, 1% by IUD, and fewer than 0.5% by use of the condom. By contrast, modern contraception represents only 73% of total contraception [End Page 44] in Pakistan and 71% in Sri Lanka. However, the proportions of women using modern contraception are highest in India, Bangladesh and Sri Lanka, which together account for nearly 50% of all the women concerned.
The condom is not widely used in this region. The highest proportion of users (5.5% of all women of childbearing age in a union) is observed in Pakistan. Use of the IUD is equally limited, with the highest proportion (5.1%) observed in Sri Lanka. The pill is popular only in Bangladesh, where it is used by 26% of all women in a union, representing 45% of women using some method of contraception. In the Maldives, where contraceptive prevalence is 42%, its use is only half as frequent, and Sri Lanka, the third country where this method is in use, has only 6.7% of users. Voluntary sterilization, mainly female, is widely employed by couples in South Asia for limiting family size. Among women of childbearing ages in the Maldives, 10% are protected by female sterilization, and the proportions are 15% in Nepal, 21% in Sri Lanka and 34.2% in India. In South Asia, as elsewhere in the world, there is a clear inverse relationship between fertility and contraception (Figure 10).
Sterilization is the main birth control method in India
Attitudes towards female and male sterilization are extremely diverse in this world region. Sterilization is non-existent in Afghanistan, and is not widely practised in Bhutan, Bangladesh and Pakistan (of these the maximum is in Pakistan, where 7% of women in a union are protected by female sterilization). As mentioned, the proportion of women aged 15-49 relying on sterilization to [End Page 45] regulate their fertility is substantially higher elsewhere, but India stands out from the other countries, since over a third of all women are concerned(26). Male sterilization for birth control purposes, on the other hand, is everywhere at a low level, with the exception of Bhutan where 8% of women in a union are protected by the sterilization of their spouse or partner and 3% by their own sterilization.
Recourse to sterilization, and essentially female sterilization, has been widespread in India since the mid 1980s (Rajan and Véron, 2006). In 2005-2006, 66% of married women aged 15-49 had been sterilized, and the proportion was 77% among those using modern contraception, whereas only 1% were protected by sterilization of their husband. Trends over the last fifteen years in contraception and in the relative importance of sterilization can be followed using the three NFHS surveys conducted in India (Table 17). Between 1992-1993 and 2005-2006, contraceptive prevalence increased by 15 percentage points, taking to over one half the proportion of women of childbearing age and in a union who now use contraception, against 41% at the start of the 1990s. Over a long period, sterilization and in particular male sterilization was encouraged in India(27), [End Page 46] until the excesses committed during the state of emergency (1975-1977) led the central government to issue statements in favour of reversible methods of contraception. In spite of this, female sterilization remains by far the most common means used by Indian couples for controlling their fertility, and even increased by a further 10 percentage points between 1992-1993 and 2005-2006. Official pronouncements intended to promote male rather than female sterilization have had little effect. The proportion of couples relying on male sterilization has remained small and has actually declined over the last fifteen years. All in all, couples judge female sterilization to be the simplest means of controlling their fertility: 77% of sterilized women have chosen this option directly without previously using another method of birth control. Age at sterilization is tending to decrease: the median age of women at sterilization fell from 26.6 in 1992-1993 to 25.5 in 2005-2006. Of 100 sterilized women in the most recent survey, 38 were sterilized at age 20-25 and 35 at age 25-30.
Contraceptive methods used in India in the 1990s and 2000s (prevalence in %)
|Contraceptive method used||1992-1993||1998-1999||2005-2006|
|Population: women aged 15-49 living in union.|
|Sources: NFHS, NFHS-2 and NFHS-3.|
Country-wide data on sterilization in India conceal large regional disparities. Only 8% of couples are protected by female sterilization in the state of Manipur against 63% in Andhra Pradesh; and whereas women's median age at sterilization is 30 in the former, it is a little over 23 in the latter.
X. Overall mortality decline, but widely separated extremes
A general improvement in life expectancy
As in the rest of the world, mortality has declined in all countries of the region since 1950 (Figure 11 and Table 18). The region's two extremes are Afghanistan, where life expectancy at birth is among the lowest in the world (42 years in 2000-2005), and Sri Lanka, where the mean length of life, close to 71 years in 2000-2005, (i.e. nearly 30 years longer) is above the world average. [End Page 47] The other countries are in intermediate positions. Although the mortality transitions in these countries differ in some respects, the increases in life expectancy at birth are closely comparable, and in 2000-2005 the populations of Bangladesh, Bhutan, India, the Maldives, Nepal and Pakistan all had lengths of life between 61 and 66 years. These countries are closing the gap with Sri Lanka, where the rate of progress is slowing somewhat, although in 2000-2005, life expectancy there was still five years longer than in the Maldives, the top-ranking intermediate group country.
Gender differences in mortality have also evolved substantially over the last fifty years. In the early 1950s, life expectancy at birth was lower for women than for men in every country of the region except Bhutan, as is shown by the index based on the female/male ratio of life expectancies at birth (Figure 12). This result was in direct contradiction with the biological norms that are responsible for men having higher mortality than women at all ages and that, in the absence of sex discrimination, lead to a shorter life expectancy for males than for females. In 1950-1955, this index stood at the particularly low level of 94 in the Maldives and in Pakistan, indicating a large female mortality disadvantage. The male advantage has disappeared in every country over the last fifty years. In 2000-2005, the life expectancies at birth of men and women in Afghanistan and the Maldives were identical. Elsewhere, women are living longer than men. The most favourable relative situation for women is in Sri Lanka, where the ratio is 112 (Figure 12). In the past, the female life expectancy [End Page 48] disadvantage was explained in part by maternal mortality, but this was not the only factor involved. Life tables for the years 1945-1947 in Sri Lanka, and for 1970-1972 in India, show an excess female mortality at all ages from birth up to around age 40 (ESCAP, 1982) due to sex inequality in levels of care, attention and education. Discrimination against girls still exists, particularly in certain states of India. It occurs chiefly before birth and, in the countries where the situation in this respect is least favourable, women's life expectancy is at minimum equal to men's. Levels of maternal mortality still diverge widely today: the number of maternal deaths per 100,000 births is 92 in Sri Lanka and estimated at 1,900 in Afghanistan (Table 18).
Life expectancy since the 1950s, maternal mortality in 2000 and HIV prevalence in 2005
|Country||Life expectancy (in years)||Maternal
|Sources: United Nations (2007a), WHO and UNAIDS.|
|Sri Lanka||57.6||62.2||65.0||68.8||70.4||70.8||92||< 0.1|
Although the numbers of HIV-positive persons can be large because of population size – in India especially(28) – prevalence rates for HIV/AIDS remain relatively low in the region. The maximum is observed in Nepal where the rate among persons aged 15-49 is 0.5% compared with a world average of 0.8%(29). The prevalence rate of 0.9% initially given for India in 2005 was subsequently revised downwards on the basis of the results from NFHS-3(30) which found a prevalence rate of 0.28% in the 15-49 age group. The epidemic affects men more than women, and urban more than rural populations (Table 19). It also affects [End Page 49]
the states of India in different ways, as shown by the regional variations in prevalence (men and women combined): 0.07% in Uttar Pradesh, 0.34% in Tamil Nadu, 0.62 in Maharahtra, 0.69% in Karnataka, 0.97% in Andhra Pradesh and 1.13% in Manipur (prevalence in the other states is 0.13%)(31). Among women, the prevalence rate is lower for those with no education than for those with at least five years of education (0.27% and 0.49% respectively) but then declines as educational level rises (0.07% for women with at least 12 years in education). Among men, the rate is highest for those with no education (0.5%). It is lower
HIV/AIDS prevalence rate in the adult population (aged 15-49) in India, 2005-2006 (in %)
|Source: NFHS-3 survey.|
[End Page 50]
for those with at least five years of education (0.36%) then rises again for those with between five and seven years of education and subsequently falls with length of education (the prevalence rate is 0.16% for men with more than 12 years in education). We note that the highest prevalence rates are observed for women who had sexual intercourse with a high-risk partner in the 12 months before the survey (rate 2.23%) and for those who did not use a condom (rate 2.83%). Prevalence rates for men in the same categories are considerably lower.
For a number of countries in South Asia, the 2005 United Nations report on mortality gives values for male and female life expectancy at birth and at ages 15 and 60 at two separate periods in time (Table 20). The decline in
Male and female life expectancy at birth, at age 15 and at age 60 (in years)
|Source: United Nations (2006b).|
[End Page 51]
mortality at younger ages accounts for some of the rise in life expectancy at birth but is not the only factor involved since life expectancies at age 15 are also rising. Life expectancies at age 60 are following a more disparate course though the trend is upward.
Early-age mortality remains high in most countries
Infant mortality fell throughout the region between 1960-1965 and 2000 2005 (Table 21 and Figure 13). Sri Lanka and Afghanistan are again in the extreme positions: by the 1960s, infant mortality in Sri Lanka was already
Infant mortality since the early 1950s
|Infant mortality rate (‰)||Change (%)|
|Source: United Nations (2007a).|
[End Page 52]
much lower than in most countries of South Asia, while in Afghanistan infant mortality was among the highest in the world. Considerable progress has been achieved in the other countries of the region, though infant mortality was still relatively high in the early 2000s: 46‰ in the Maldives, 53‰ in Bhutan, 61‰ in Bangladesh, 63‰ in India, 65‰ in Nepal and 75‰ in Pakistan.
According to United Nations estimates (2007a), under-five mortality in Afghanistan has not decreased over the last ten years: it was 257‰ in 1990 1995 and 252‰ in 2000-2005. At 14‰ in 2000-2005, under-five mortality in Sri Lanka bears no comparison with that in Afghanistan (Table 22). The next best level of under-five mortality is in the Maldives, where the rate declined considerably in the 1990s but still stood at 59‰ in 2000-2005. Infant and child mortality levels in the other countries were very similar in 1990-1995, with all rates situated between 124‰ and 128‰, but ten years later had become more disparate, ranging from 108‰ in Pakistan to 78‰ in Bhutan.
Child health is closely correlated with mother's level of education
Child mortality levels reflect the persistent disparities in child health in South Asia. The proportions of deliveries attended by skilled health personnel, the percentage of malnourished children and the levels of child vaccination coverage provide further proof of this diversity (Table 22).
Child mortality, delivery conditions, malnutrition and vaccination
|Country||Mortality rate (‰)||Deliveries
|(a) Children receiving the full set of vaccinations (BCG, DPT, polio and measles).|
|(b) Children under five with stunted growth.|
|Sources: United Nations, World Health Statistics 2007, DHS surveys and United Nations (2007a).|
|Afghanistan||60||168||252||14 (2003)||53.6 (1997)||–|
|Bangladesh||36||61||83||13 (2004)||50.5 (2004)||73.1 (2004)|
|India||39||62||90||48 (2006)||51.0 (1998/99)||42.0 (1999)|
|Maldives||24||46||59||70 (2001)||31.9 (2001)||–|
|Nepal||32||64||88||19 (2006)||57.1 (2001)||82.8 (2006)|
|Pakistan||53||75||108||31 (2005)||41.5 (2001)||35.1 (1991)|
|Sri Lanka||8||12||14||97 (2000)||18.4 (2000)||67.3 (1987)|
[End Page 53]
Child mortality, delivery conditions and vaccination by area of residence and educational level of mother, India, 2005-2006
|Mother's level of education
(years of shooling)
|Infant mortality (‰)||57.0||41.5||62.2||69.7||66.0||49.5||41.5||36.5||25.9|
|Mortality 1-5 years (‰)||18.4||10.6||21.0||29.9||13.8||11.5||5.6||3.6||3.9|
|Mortality 0-5 years (‰)||74.3||51.7||82.0||94.7||78.8||60.5||46.9||40.0||29.7|
by trained personnel (%)
|Children aged 1 or 2
receiving the main
The conditions of birth vary widely. In Sri Lanka, 97% of deliveries are attended by skilled health personnel, against 70% in the Maldives, around 50% in Bhutan and India, 14% in Afghanistan and 13% in Bangladesh. The proportion of children vaccinated is also an indicator of health conditions. Though the relevant data are not available for all the countries of the region and some are out of date, in the 2000s the proportion of children vaccinated against the main diseases exceeds 70% in Bangladesh and 80% in Nepal(32).
Data for India on infant mortality in 2005-2006 highlight the impact of area of residence and mother's level of education on child health (Table 23). Infant mortality, which has an average value for India of 57‰ in 2005-2006, varies by 20 points depending on whether the area of residence is urban or rural. For children under one, the mortality rate varies by 44 points according to whether the mothers have received no education (70‰) or more than 12 years of education (26‰). In India as elsewhere in the world, the mortality differentials are even larger at ages 1-5. The conditions of delivery are much more favourable in urban areas than in rural areas, since well over 70% of urban births are attended by trained health personnel, compared with less than 40% of rural births. Similarly, the conditions of birth vary considerably with the mother's educational level, ranging from 26% of deliveries attended by trained personnel for women with no education to 91% for women with at least 12 years of education. The percentage of children receiving the basic vaccinations varies by almost 20 points depending on whether the mother lives in an urban or rural area, and by nearly 50 points between the least and most educated women. [End Page 54]
Health inequalities thus remain extremely large in South Asia, whether comparison is between countries or between socioeconomic strata in the population of a single country. India introduced a new health policy in 2002(33). Among its clearly stated objectives, based on the principle of equity, is that of reducing the inequalities in access to healthcare associated with urban versus rural residence or with differences in the development level of states or in the economic status of populations. However, it will be some years before its effects become visible.
XI. Age structures: demographic windows of opportunity
Median age, a crude indicator of age structure, reveals three sharply contrasting situations inherited from past trends in mortality and fertility (Table 24). A contrast is again observed between Afghanistan, where the fertility transition has hardly started and the population is particularly young (median age 16.4), and Sri Lanka, where the transition is complete and the population is much older (median age 29.5). In the other countries, all of which are engaged in the fertility transition, median ages range from 20 to 24.
In 2005, all the countries of South Asia with the exception of Sri Lanka had populations with large proportions of young people. Except in Sri Lanka, where the proportion was 24%, the under-15s make up over a third of the population (Table 23). The proportion of young people under 15 is around 40% in Nepal and reaches 47% in Afghanistan.
Indicators of age-sex structure in the countries of South Asia, 2005
|Countries||Median age (years)||Under-15s (%)||Over-60s (%)||Sex ratio (%)|
|Source: United Nations (2007a).|
[End Page 55]
Using the proportion – not the absolute number – of people over 60 as the criterion of population ageing, it is clear that the process has made little impact so far in the countries of the region. Only in Sri Lanka is the proportion of over-60s close to 10%, as a consequence of the longstanding fertility decline, whereas it is 7.5% in India and below 7% in the other countries.
Population projections to 2020 show that Sri Lanka will see a marked ageing of its population in the years ahead, with the proportion of over-60s exceeding 17% by that year (Appendix Table A.11). The next "oldest" country will be India, with 10% of its population over 60. Everywhere else, the over 60s will still represent less – in some cases much less – than one-tenth of the population. The proportion of persons aged 60 and over is currently very low in Afghanistan (4%) and is unlikely to change between now and 2020.
The populations in the countries of South Asia are also differentiated by their sex ratios. The largest male surpluses are found in Bhutan, Afghanistan and India; men are in the minority in Nepal and Sri Lanka, where there were 98 males for 100 females in 2005.
The large sex imbalance at birth affecting several East and Southeast Asian countries, among them China, is not observed in this region except in India, and then only in the country's north-west states. We will return later to this sex imbalance that reflects strong discrimination against girl children in these states.
Age distributions reflect the progress of the demographic transition
The stage of demographic transition reached by the countries of the region is illustrated by their age structures (Figures 14A-14H). The fertility decline in Sri Lanka began many years ago, as shown by the narrower base of the pyramid starting from the 35-39 age group. Despite an "echo" effect(34), cohort size has fallen continuously since 1980-1985. Compared with Sri Lanka, the fertility transition has come later in Bhutan and the Maldives, but cohort size has been falling in both countries for the last twenty and ten years respectively. The base of the population pyramid for Pakistan is starting to contract, while the pyramids for Bangladesh, India and Nepal are very similar, with as yet no major decrease in numbers at young ages. Despite a decline in fertility, the numbers of births are not yet falling, because the population of childbearing age is still large. Afghanistan, finally, presents an age structure characteristic of a country with high fertility and high mortality.
Between 2005 and 2020, the proportion of people aged 15-60, i.e. in the working-age population, will increase in all the countries (except Afghanistan). [End Page 56]
[End Page 57]
Further increase can be expected even in countries where the proportion of people aged 15-60 was already high in 2005, like Bhutan and the Maldives (60%). This relative increase in the working-age population, before the reduction in the economically inactive young population is cancelled out by the increase in the inactive older population, constitutes the "demographic window". At this point countries have an opportunity for development, since the dependency ratio is favourable to wealth creation, thanks to the reduced public burden of education costs for the young and of health and pension costs for older people. In Bhutan, people aged 15-60 will represent 67% of the population in 2020. For the countries in a position to benefit, the demographic window represents an opportunity. But the advantage is a potential one, and for it to be realized, the working-age population must be employed in the full sense of the term, with low under-employment or unemployment. Sri Lanka differs from the other countries in that it is already benefiting from the demographic window (the proportion aged 15-60 is currently 66% and will probably descend to 62% by 2020).
XII. A relatively low urban-rural ratio
Moderate urbanization but strong urban growth
In South Asia, like elsewhere in the world, the criteria used for classifying populations as "urban" are not all the same. The definition of urban in India can be administrative (localities possessing a municipal charter, etc.) or based on a combination of criteria: number of inhabitants, population density, proportion of men in non-agricultural occupations. In neighbouring Sri Lanka, only populations in "municipal zones and zones with an urban council" are classified as urban (Table 25). In Pakistan, an urban settlement is also defined
Criteria for defining urban populations in four countries of South Asia
|Country||Definition of "urban"|
|Source: United Nations(2007b).|
|India||Towns and cities: localities possessing a municipal charter, a committee for a municipal
zone or designated urban zone or grouped zone (cantonment) or localities with
a population of 5,000 or more inhabitants, a population density of 1,000 or more
inhabitants per square mile or 400 per sq.km, definite urban characteristics,
and with three-quarters of adult males in non-agricultural employment.
|Pakistan||Localities possessing a municipal charter or a municipal committee and grouped zones
|Sri Lanka||Urban sector: comprising all the municipal zones and zones possessing an urban council.|
[End Page 58]
in uniquely administrative terms. In Nepal, localities needed 5,000 inhabitants to be classified as urban in the 1961 census and 9,000 in subsequent censuses. The criteria for population size and density, continuous built-up area and occupational structure have all changed over time, and since 1999 the definition based on population size, annual income and infrastructure varies between ecological zones according to complex rules (United Nations, 2008)(35). The definition of urban in Bangladesh is similar to that in India but more complex since it also includes public amenities, access to water, educational level, etc.
The countries of South Asia are still not heavily urbanized (Table 26). In Pakistan, the country with the region's highest percentage urban in 2005, only 35% of the population lived in towns or cities. The lowest levels of urbanization are observed in Sri Lanka and Nepal, where barely 15% of the inhabitants are urban dwellers. In India, less than 30% of the population today lives in towns or cities, which is three times more than in 1901 and twice the level in 1941, but still a small proportion compared with the world average (50%).
Percentage urban, 1950-2005, and projections to 2030
|Source: United Nations (2008).|
Although the proportion of people living in towns and cities is rising at a moderate pace, the total urban population in South Asia has grown sharply over fifty years. It numbered 73.7 million in 1950 and 433.2 million in 2005, representing nearly a six-fold increase (Figure 15). India accounted for much of this urban growth, since of the 433.2 million urban dwellers in South Asia in 2005, three-quarters lived in India. The pace of change in the urban population differs between countries, and in some instances varies greatly between periods within the same country. In Afghanistan, in particular, the urban population grew at an annual rate of 6% in the period 2000-2005, but the rate was negative in the period 1980-1985(36) then reached 8.5% in the years 1990-1995. For the [End Page 59] urban population of Bangladesh, the rate is currently 3.5%, but was in excess of 10% between 1975 and 1980. In Bhutan, the rate of urban population growth since 1950 has fluctuated around an annual average of 5%. The urban growth rate in India climbed from 2.6% in 1950-1955 to 3.8% in 1970-1975 then slowed to its present level of 2.3%. Like Bangladesh, the Maldives experienced a period of rapid urban growth, with a rate of 10% in 1970-1975. Between 1970 and 2000, the urban population in Nepal grew by over 6% a year on average, whereas in Pakistan the rate was between 4% and 4.5% over several decades. Again the pattern for Sri Lanka is singular: the growth rate rose from 2.8% in 1950-1955 to 4.3% in 1965-1970, then fell back to below 1% and is currently close to zero (0.15% in 2000-2005).
According to United Nations forecasts, Pakistan will remain the most urbanized country in the region up to 2030, with nearly one in two inhabitants living in urban areas. Urbanization in Sri Lanka, meanwhile, at slightly over 20%, will still be remarkably low. In Bangladesh and India, 40% of the population will be urban. Although this level might be considered moderate in 2030, it represents a total of 590 million urban-dwellers in India alone.
Five of the world's thirteen largest cities are in South Asia
The countries of South Asia are very different in size, and their capital cities likewise. Delhi had 15 million inhabitants in 2005, Katmandu well below one million, and Male, capital of the Maldives, only 89,000 (Table 27).
The city with the most spectacular growth over the last fifty years is Dhaka, whose population increased thirty-fold in the period 1950-2005. The city [End Page 60]
Population change in selected capital and largest cities between 1950 and 2000 and forecasts for 2015
|Country||Cities||Population (in thousands)||Ratio||Percentage of total
|Source: United Nations, World Urbanization Prospects: the 2005 Revision.|
[End Page 61]
became the capital of the newly created Bangladesh in 1971. Between 1975 and 2005, the seven-fold increase in the population of Islamabad was faster than that of Dhaka, where population increased six-fold. The populations of Kabul and Katmandu grew by a factor of 4.4-4.5 over the same period. In fifty years, the population of Delhi has grown more than ten-fold, increasing from 1.4 million in 1950 to 15 million in 2005.
The demographic weight of the capital cities in their respective countries also varies. Delhi is home to less than 2% of the population of India, whereas 10% of that of Afghanistan lives in Kabul. The inhabitants of the Afghan capital represent 44% of that country's urban population, whereas the population of Kathmandu represents less than 20% of Nepal's urban population and that of Delhi less than 5% of India's.
The urbanization process is singular in India on account of the population size and its rate of growth in the twentieth century. A long-term comparison with the United States has shown that not only is the percentage urban lower in India but it is not catching up with the American level, even though India has a larger urban population than the United States (Véron, 1987). India already had a number of large cities at the end of the nineteenth century: in 1891, 744,000 people lived in Calcutta, 822,000 in Bombay, 453,000 in Madras and 415,000 in Hyderabad (Davis, 1951), In the 2001 census, India counted 34 urban agglomerations with over one million inhabitants and six with over five million inhabitants: Mumbai (formerly Bombay, 16.4 million), Kolkata (Calcutta, 13.2 million), Delhi (12.9 million), Chennai (formerly Madras, 6.6 million), Hyderabad and Bangalore (each with 5.7 million).
On a global level, of 21 urban agglomerations with over 10 million inhabitants in 2005, five of them – not all capital cities – were in South Asia: Dhaka, Delhi, Kolkata, Mumbai and Karachi. These five urban agglomerations are among the 13 largest in the world.
XIII. A lack of data on international migration
The countries of South Asia are emigration countries. Three types of migration can be identified: migration leading to permanent settlement in Europe, Australia and North America; labour migration to countries in the Middle East; and seasonal migration, mostly localized in border regions, like that into India from Bangladesh and Nepal (IOM, 2005). The largest migration flows are to the Gulf countries. When no visa is required to enter a neighbouring country – as is the case between India and Nepal, between India and Bhutan and between Bhutan and Nepal (Khadria, 2005) – cross-border migration is indistinct from internal migration. Where a visa is required, as is the case for Bangladesh and India, a proportion of this migration is illegal. This local migration involves primarily the poorest populations, such as the farmers from the Nepal valleys who go to north-east India at harvest time. [End Page 62]
In the last few years, the range of destination countries for South Asian migrants has broadened (Khadria, 2005). Increasingly, migrants are attracted to Japan, Hong Kong, South Korea, Malaysia, Singapore and Taiwan. Examples include plantation workers in Malaysia (many Bangladeshis), domestic staff in Singapore and construction workers in South Korea (Nepalese in particular). Migrants include large numbers of women, mainly in the domestic service sector. Highly qualified migrants, meanwhile, leave for Europe, North America or Australia.
Afghanistan experienced large-scale population movements during the war with the Soviet Union between 1979 and 1989. All three main types of migration – corresponding to permanent settlement in Europe, North America or Australia; labour migration to the Middle East and certain South-East Asian countries; and cross-border moves, such as migration to India – occur in Bangladesh though the respective scale of these flows is not known. Pakistanis also leave their country to settle in Europe, North America or East Asia, entering either legally, or as students (some of whom later seek to stay on), or, for the poorest, illegally. Sri Lanka too is experiencing large-scale international migration. Political reasons are causing Tamils to leave the country and seek political refugee status in the West (Etiemble, 2004). The Gulf countries remain a major destination for migrants from Sri Lanka. Women account for 70% of this migration, reflecting the high demand in these countries for female domestic workers.
Migration is a growing phenomenon in India, though its in-depth study is limited essentially to Kerala. A survey from the late 1990s has been used to analyse the impact of migration on population age structure, family life (temporary separation of spouses), average educational level (through the selectivity of migration), poverty, unemployment, etc. (Zachariah et al., 2003). It emerges, for example, that women with a migrant husband in a Gulf country have high status because they receive remittances in their own name but are weighed down by increased responsibilities due to their husband's absence, and by loneliness: 60% would like their husband to come home. Return migration is becoming much more common, raising the issue of reintegrating migrants into the society of origin. Forty percent of return migrants set up their own business, but a proportion of returnees have great difficulty in reintegrating Kerala society and, though still of working age, do not find employment and use their savings to maintain a high standard of living (Zachariah et al., 2002; Zachariah et al., 2006).
The economic importance of international migration for each country can be measured from the financial transfers made by its citizens resident abroad. In 2004, migration-related transfers represented a varying but in some cases large proportion of gross national income in the countries of South Asia: 12.1% in Nepal, 8% in Sri Lanka, 5.5% in Bangladesh, 4.2% in Pakistan, 3.2% in India and 0.4% in the Maldives (United Nations, 2006c). [End Page 63]
Sending countries can also be receiving countries. The United Nations estimates that 191 million people worldwide lived outside their country of birth in 2005, representing 3% of the world population(37) (United Nations, 2006c). The South Asian country with the largest number of foreign-born persons is India, where they total 3.7 million(38). This migrant stock is the eighth largest in the world, though it represents a mere 0.5% of India's total population (Appendix Table A.12). Pakistan currently has some 3.3 million foreign-born residents, equivalent to 2.1% of the total population. The migrant stock thus defined is 1 million in Bangladesh (0.7% of the population), 819,000 in Nepal (3%), 368,000 in Sri Lanka (1.8%), 43,000 in Afghanistan (0.1%), 10,000 in Bhutan (0.5%) and 3,000 in the Maldives (1%).
XIV. Access to education remains unequal
Access to education plays a large role in the population dynamics of all countries, affecting factors such as the pace of demographic transition, contraceptive practice or child health.
The countries of South Asia present contrasting situations in terms of literacy. In the early 2000s, literacy among adults over age 15 was above 90% in Sri Lanka and 96% in the Maldives (Table 28). The contrast with Afghanistan, where only 28% of the adult population is literate, is particularly sharp. Literacy levels are around 50% in Bangladesh, Nepal and Pakistan and over 60% in India.
Overlaying these differences in literacy levels are varying degrees of gender inequality (Figure 16). The situation in the Maldives and Sri Lanka is one of
Literacy rates of adults aged 15 and over in the 2000s
|Sri Lanka (2001)||90.7|
[End Page 64]
high equality, with literacy levels at around 90% for men and women alike. In Nepal and Pakistan, on the other hand, inequality is marked and the female adult literacy rate is less than 60% of the male adult rate. This form of inequality is greater still in Afghanistan where female literacy rates represented 29% of the male rate in 2000.
The population of India has a higher educational level than neighbouring countries like Bangladesh and Pakistan, but in this domain, as elsewhere, the country has extreme regional disparities (Banthia, 2001). In 2001, the average literacy rate of the population aged 7 and over was 65%, but stood at over 90% in Kerala and below 48% in Bihar. For men, the average literacy rate is 76% and varies between a maximum of 94% in Kerala and a minimum of 60% in Bihar, while for women the average is 54% with a maximum of 88% in Kerala and a minimum of 34% in Bihar.
XV. Increasing vulnerability
In some countries of the region, violent conflict or extreme tensions have created a climate of insecurity that is unfavourable to development at the present time. However, increased vulnerability of the populations of South Asia in economic, social and ecological terms is also a legacy of rapid population growth in the past.
Economic and social vulnerability
Several populations are particularly vulnerable today in economic and social terms. In some regions of India, strong discrimination against girl [End Page 65] children is reflected in sex-selective abortion practices. Results from the 2001 Indian census showing a high sex ratio in the young population revealed the extent of discrimination against girl children in the north-west of the country(39). In the Punjab and in Haryana, which count among India's most developed states, this discrimination was such that the sex ratio in the population aged 0-6 reached 130 boys for 100 girls in some districts and even higher levels in the district of Fatehgarh Sahib (Nanda and Véron, 2004). Women's educational level is relatively high in the Punjab (the female literacy rate was 64% in 2001), yet female foetuses are nonetheless aborted in large numbers, the result of a traditional boy preference that may have been rekindled by falling birth rates(40). In recent years, the government has made efforts to stamp out the practice of sex-selective abortion via information campaigns and sanctions against practitioners who tell couples the sex of an unborn child, but without much apparent success. For India as a whole, the sex ratio of the 0-6 age group was 108.9 in the 2005-2006 NFHS, against 107.9 in the 2001 census. The sex ratio has not changed in urban areas but has risen in rural areas.
Vulnerability also takes other forms(41). Vulnerable populations include child workers, unemployed adults, and those, particularly at older ages, without access to a system of social protection. Although the demographic weight of older people remains limited in South Asia, their number is rising and, in India for example, the question of population ageing is an emerging concern (Bose and Shankardass, 2004). The country today has more than 80 million people over age 60 and by 2016 is likely to have more than 110 million. Since half of all men over age 60, and more than 70% of women, are economically dependent on other people, their economic security is set to become a crucial issue.
Another form of vulnerability is linked to the problem of urban housing among the poorest populations as evidenced by the spontaneous settlements and shanty towns that spring up on city outskirts. UN-Habitat estimates that over 230 million people lived in shanty towns in South Asia in 2001(42). According to the same source, shanty town-dwellers numbered 30 million in Bangladesh, 36 million in Pakistan and 158 million in India. In addition to economic and social vulnerability, the populations may also face ecological vulnerability. [End Page 66]
Living conditions for a large proportion of the population of Bangladesh in the Ganges-Brahmaputra delta are particularly hazardous. High population densities mean that each episode of flooding produces large numbers of victims. The history of natural disasters in Bangladesh over the last two hundred years reveals the full extent of the country's exposure to ecological hazards (Table 29). In late 2007, the agricultural sector was again badly hit by a tropical cyclone, with losses to nearly 7 million people.
Major disasters in Bangladesh over two centuries
|Source: W. Brian Arthur and GeoffreyMcNicoll, "An analytical survey of population and development in Bangladesh", Population and Development Review, 4(1), March 1978, p. 29,cited by Escap (1981).|
|1769-1776||Great famine of Bengal||Nearly one-third reduction in theBengal population,
though impact was lighter in West Bengal
|1784-1788||Floods and famine.Brahmaputra
River changed its course
and tidal wave
|Around 400,000 deaths|
|1897||Chittagong cyclone||Around 175,000 deaths|
|1918-1919||Spanish influenza||Around 400,000 deaths|
|1943||Bengal famine||Between 2 and 2.5 million deaths|
|1947||Partition of India||Unknown. Total deaths in partition numbered
1 million, but most occurred in the West
|1970||Cyclone and tidal wave||Between 200,000 and 500,000 deaths|
|1971||War of Independence||Around 500,000 deaths|
|1974||Famine||Official death toll 30,000, but some estimates are much higher (including reference to 500,000 deaths with 80,000 for the Rangpur district alone).|
In other countries of the region, such as India, the Maldives or Sri Lanka, substantial proportions of the population are severely exposed to rising water levels. The earlier period of rapid population growth in these countries brought a large increase in population densities. In India, average population density rose from 77 inhabitants per sq.km in 1901 to 324 in 2001; it currently stands at over 800 inhabitants per sq.km in Kerala and 900 in West Bengal. The resulting pressure on land has led to increased settlement in coastal zones, which are barely above sea level. The consequences of the 2004 tsunami were particularly disastrous in Sri Lanka, in the Indian states of Tamil Nadu and Kerala, and in the Maldives, all of which had large populations living at sea level. In India, it left 11,000 persons dead, 5,600 missing and around 7,000 injured. The worst affected Indian state, Tamil Nadu, alone counted [End Page 67] 8,000 dead(43). In Sri Lanka, the tsunami caused 40,000 deaths and displaced 2.5 million persons. Rising water levels due to climate change threaten a growing number of people in Bangladesh, India, the Maldives and Sri Lanka.
Today home to a population of around 1.6 billion people, South Asia is a region of contrasts in both its geography and demography. Where Afghanistan is composed mainly of mountains and includes numerous high peaks, most of Sri Lanka is occupied by plains. It is between these two countries also that the demographic contrasts are sharpest. Mortality in Afghanistan, though it has fallen, is still extremely high, since life expectancy at birth scarcely exceeds 40 years, whereas in Sri Lanka mean length of life was over 70 years in 2000-2005. And whereas demographic transition has not begun in Afghanistan, in Sri Lanka it is complete. One consequence of these changes in mortality and fertility is that the population of Afghanistan is still extremely young (median age 16 in 2005) while that of Sri Lanka is much older (median age close to 30). The other countries of the region – Bangladesh, Bhutan, India, the Maldives, Nepal and Pakistan – are in intermediate positions for all demographic indicators. In all of them the demographic transition is under way. Mortality has fallen appreciably and life expectancy at birth is 62 years minimum. Fertility has fallen everywhere though large differences remain: women in Bhutan and the Maldives produce fewer than three children whereas the figure is still four in Pakistan.
From a demographic point of view, South Asia is heavily dominated by India, whose territory contains more than two-thirds of the region's population. After a four-fold increase in its population over a hundred years, India sought to limit population growth through birth control programmes from the early 1950s, and today the country's demographic transition is well advanced. But there are large regional disparities. The National Family and Health Survey of 2005-2006 shows that infant mortality – average 57‰ for the country as a whole – ranges by state between 15‰ (Kerala and Goa) and 71‰ (Chhattisgarh). On fertility, the average for India is 2.1 children per woman, and the variation is between 1.8 (Andhra Pradesh and Goa) and 4 (Bihar).
In terms of development, the different countries face broadly the same challenges: finding employment for large working-age populations, reducing poverty, creating a system of social protection for the poor and the old, and ending the various forms of discrimination against women. Other challenges are more specific to particular countries. Bangladesh, Pakistan and India must find ways to sustain the growth of urban agglomerations with over [End Page 68] 10 million inhabitants, The same is true for access to education: in Sri Lanka and the Maldives this is barely an issue – literacy levels are above 90% – but the question of education is important in Bangladesh and Pakistan (where literacy rates are only 48% and 50%) and fundamental in Afghanistan (where literacy was only 28% in 2000). These countries also face the challenge of reducing gender inequalities in access to education.
By 2040, South Asia can expect to have 2.2 billion inhabitants, representing an additional 600 million inhabitants with respect to its present population. Immense challenges will need to be overcome if the countries of this region are to achieve even a modest degree of sustainable development.
The author would like to thank Cédric Yaël Mathieu for his valuable and effective assistance in compiling the database. [End Page 69]
Censuses, national fertility and health surveys and other national demographic household surveys in South Asia from 1960 to 2007
|* Bangladesh Maternal Health Services and Maternal Mortality Survey 2001.|
|** Surveys posted on the website of the Department of Census and Statistics, Sri Lanka, but not on the Macro International website.|
|Sources: www.measuredhs.com, http://surveynetwork.org|
|Afghanistan||1979||–||2000, 2003, 2005|
|Bangladesh||1974, 1981, 1991, 2001||1993/94, 1996-1997, 1999-
2000, 2001*, 2004, 2007
|1975, 1995, 2000, 2001,
|India||1961, 1971, 1981, 1991,
|2000, 2001, 2002, 2006|
|Maldives||1977, 1985, 1990, 1995,
|Nepal||1961, 1971, 1981, 1991,
|1987, 1996, 2001, 2006||1976, 1981, 1986, 2002|
|Pakistan||1961, 1972, 1981, 1998||1990-1991, 2006||1975, 2002, 2003, 2004|
|Sri Lanka||1963, 1971, 1981, 2000||1987, 1993**, 2000**||1975, 1985, 1994, 1999,
2000, 2002, 2003
[End Page 70]
Land area and densities in 2000, and population of the countries of South Asia, 1950-2040
sq. km of
|Number of inhabitants (in thousands)|
|Sources: United Nations Common Database (http://unstats.un.org/unsd/cdb/) and World Population Prospects: The 2006 Revision (http://www.un.org/esa/population/publications/wpp2006/wpp2006.htm); FAO Statistical Yearbook, 2005-2006 for arable land; Pison (2007), "The population of the world", Population & Societies, 436.|
|Sri Lanka||United Kingdom||1948||66||285||2,091||7,339||12,342||17,114||18,714||19,576||19,671|
[End Page 71]
Birth rate, death rate and total growth rate in the countries of South Asia, 1950-2005
|Country||Crude birth rate (‰)||Crude death rate (‰)||Mean annual growth rate (%)|
|Source: United Nations (2007a).|
[End Page 72]
Median age at first marriage (men and women), prevalence of polygyny and proportion of never-married women in the countries of South Asia
|Country||Date of latest
|Median age at first
|Percentage of women
in polygamous unions
|Men||Women||age 15-49||age 35-44|
|* Analysis in progress.|
|Source: DHS surveys.|
|India||2005-2006||23.4||17.2||6.2||0.6 (age 45-49)|
|Nepal||2006||20.6||17.2||3.4||4.4||6.2||1.2 (age 45-49)|
[End Page 73]
Total fertility rate and net reproduction rate in the countries of South Asia
|Country||Total fertility rate(a)||Net reproduction rate(b)|
|(a) Sum of age-specific fertility rates observed over a given period. The TFR can be interpreted as the average number of children that a woman would bear throughout her reproductive life if she were to experience the fertility rates of the period at each age. It does not take mortality into account.|
|(b) Average number of daughters that a woman would bear throughout her reproductive life if she were to experience the fertility and mortality rates of the period at each age.|
|Source: United Nations (2007a), World Population Prospects: The 2006 Revision, http://www.un.org/esa/population/publications/wpp2006/wpp2006.htm|
[End Page 74]
Most recent indices of level, timing and characteristics of fertility in the countries of South Asia
at age 40-49
at age 25-34
women who have
who have begun
|Source: DHS surveys.|
[End Page 75]
Breastfeeding, postpartum sexual abstinence and contraception in the countries of south Asia
|Contraceptive prevalence (%)
(women aged 15-49, married or in a union)
|Sources: DHS surveys, Statcompiler; United Nations, Statistics Division (http://unstats.un.org/unsd/databases.htm).|
[End Page 76]
Life expectancy and infant mortality in the countries of South Asia, 1950-2005
|Country||Life expectancy (years)||Infant mortality rate (per thousand)|
|Source: United Nations (2007a), World Population Prospects: the 2006 Revision, http://www.un.org/esa/population/publications/wpp2006/wpp2006.htm|
[End Page 77]
Conditions of delivery, mortality and health of children in the countries of South Asia
|Country||Probability of dying
of births with at
least one antenatal
|Neonatal||Below one year
|(a) Children who have received all vaccines (BCG, DPT, polio and measles).|
|(b) Children under 5 with stunted growth.|
|Sources: United Nations, World Health Statistics 2007, (http://www.who.int/whosis/whostat2007) ; DHS surveys.|
|Afghanistan||60||165||257||52 (2003)||14 (2003)||53.6 (1997)|
|Bangladesh||36||54||73||39 (1999-2000)||13 (2004)||73.1 (2004)||50.5 (2004)|
|Bhutan||30||65||75||51 (2005)||47.7 (1999)|
|India||39||56||74||65 (1998-1999)||48 (2006)||42 (1999)||51 (1998-1999)|
|Maldives||24||33||59||98 (2000)||70 (2001)||31.9 (2001)|
|Nepal||32||56||88||49 (2001)||19 (2006)||82.8 (2006)||57.1 (2001)|
|Pakistan||53||80||100||31 (2005)||35.1 (1991)||41.5 (2001)|
|Sri Lanka||8||12||14||97 (2000)||67.3 (1987)||18.4 (2000)|
[End Page 78]
Maternal mortality and HIV prevalence at ages 15-49 in the countries of South Asia
|Country||Maternal mortality rate
(for 100,000 births)
|Prevalence rate of HIV
in 2005 in the population
aged 15-49 (%)
|Sources: WHO, World Health Report 2005, p. 245 (http://www.who.int); UNAIDS, 2006 Report on the Global AIDS Epidemic, (http://www.unaids.org).|
[End Page 79]
Proportions of the population aged under 15 and over 60, median age, dependency ratio and sex ratio in the countries of South Asia, 1950-2020
|Percentage under 15||Percentage 60+||Median age||Dependency
ratio in 2005*
|* Ratio of the population aged under 15 and over 65 to the population aged 15-65.|
|Source: United Nations (2007a), World Population Prospects: The 2006 Revision, http://www.un.org/esa/population/publications/wpp2006/wpp2006.htm|
[End Page 80]
International migration, refugees and displaced populations in the countries of South Asia
|Migrant stock(a) in 2005||Net
|* Figure corrected in accordance with the population size given by the United Nations (2006a)|
|(a) Defined as the number of people born outside the country.|
|(b) Defined as the annual number of immigrants minus the annual number of emigrants between 2000 and 2005 divided by the average total population of the country.|
|(c) Persons with refugee status under the various international conventions currently in force.|
|(d) Total number of foreign refugees, asylum seekers, return refugees and internally displaced individuals under UNHCR protection.|
|Sources: United Nations (2006), International Migration 2006; UNHCR (2006), Statistical Yearbook 2005 (http://www.unhcr.fr/cgi-bin/texis/vtx/statistics/).|
|Sri Lanka||368||1.8||– 1.6||0.1||354.8|
[End Page 81]
Development indicators in the countries of South Asia
aged 15 and
over in 2004
|Human development index(c)|
|*The United Nations (2008) give a percentage urban of 31.0% for 2005.|
|(a) United Nations calculations based on national definitions.|
|(b) Per capita gross national income calculated by the World Bank in terms of purchasing power parity (PPP).|
|(c) Composite index measuring human development, including life expectancy, adult literacy rates, school enrolment ratios and per capita GDP. The closer it is to 1, the better the situation.|
|(d) Not dated. Composite index measuring deprivations in health, education and standard of living. The closer it is to 0, the better the situation.|
|(e) Index based on the same criteria and types of measure as the HDI, but reflecting inequalities between men and women.|
|(f) UNDP has not calculated indices for this country.|
|(g) In 2003.|
|Sources: UNDP (2006), Human Development Report 2006, (http://hdr.undp.org) ; United Nations, UNstat.|
[End Page 82]
School enrolment and adult literacy by sex in the countries of South Asia
|Country||Literacy rate at age 15 and over
|Net primary school enrolment ratio
(school year ending in 2004)
|Men||Women||Overall||F/M ratio||Male||Female||Overall||F/M (%)|
|Source: Unesco (2007), EFA Global Monitoring Report.|
[End Page 83]
Indicators of nuptiality, fertility and sex preference in India, 1998-1999
|Source: NFHS-2, 1998-1999.|
|Jammu and Kashmir||18.2||2.7||29.5||56.7||32.5||2.7||38.0||2.7||83.2|
|West Bengal||16.8||2.3||36.0||43.8||33.6||2.4||20.7||3.4||90.0 [End Page 84]|
[Begin Page 85]
1. For the United Nations, South Asia is made up of the same countries excluding Afghanistan, which is considered to be in Central Asia.
2. The land surface is the difference, in each country, between the total surface area and the inland waters (the main rivers and lakes). See Pison (2007).
3. The human development index was introduced by UNDP in 1990. It is a composite index that includes life expectancy, adult literacy rates and school enrolment ratio, as well as per capita GDP. It ranges from 0 to 1.
4. The system is based on a dual data collection system: a continuous process of birth and death registration in a sample of villages and urban enumeration blocks, and an independent retrospective survey over a six-month period. Matching the data from these two sources makes it possible to distinguish events registered by both sources, from those that are partially registered or not registered by one of the sources. In these two cases, a check is carried out in the field by a visit to the households concerned. The system does not provide a means for estimating events that might have been omitted by both sources (Mari Bhat et al., 1984).
5. India is composed of 28 states and 7 union territories. Delhi, Puducherry and Chandigarh are included among the territories. Each state is administered by a Governor and a prime minister, while the territories are governed directly at federal level, represented by a Lieutenant Governor.
6. Mahayana Buddhism, or the "Great Vehicle" is a form of Buddhism that appeared in the early Christian era and reintroduces certain principles Buddha had discarded, such as salvation through devotion and ritual.
7. Around the middle of the second century BC, Indo-Aryan populations crossed Afghanistan on their way to the Indus Valley. In 637 Afghanistan was invaded by Arabs, then by Genghis Khan in the early thirteenth century and taken by Tamerlane in the fourteenth century. For several centuries it was the object of numerous disputes between Mongols and Persians.
8. Australoid peoples came from the south in the fourth century BC, then Mongols speaking a Tibeto-Burman language occupied the east and the centre of the country a few centuries later. The third migration in the second century BC was the Indo-Aryan population from the south and the west into the hills and valleys of Kathmandu (Bista, 1977).
9. East Pakistan became Bangladesh in 1971.
10. Major earthquakes may have led to a drying of the Indus valley because of the waters being diverted to the Ganges River catchment.
11. The British East India Company was a trading company that obtained the trading monopoly of the Indian Ocean from Queen Elisabeth I.
12. Up until then, India's history was also that of Pakistan and Bangladesh.
13. According to figures that are sometimes quoted but which have yet to be verified, 10 million people were displaced during Partition, and 500,000 died in the conflict.
14. Figures have been put forward of 2.5 to 3 million refugees in Pakistan and 1.5 million in Iran.
15. Three new States were created after the 1991 census, namely Uttaranchal, Jharkhand and Chhattigarh, by dividing up the highly populous Indian states of Uttar Pradesh, Bihar and Madhya Pradesh.
16. With large regional variations since the data established per district for 1971 show a mean age at first marriage for women ranging from 13 to 23 years of age (Goyal, 1988).
17. As ever, the use of regional data highlights the diversity of India: the proportion of women aged 15-19 who have had a live birth is 12.1% for India as a whole; it is only 2.1% in Himachal Pradesh and close to 21% in Jharkhand.
18. The net reproduction rate of a birth cohort is the ratio of the number of daughters born to the women of this cohort to the number of these women at birth. The period net reproduction rate is the equivalent measure calculated using cross-sectional data. A net reproduction rate of one means that the cohort exactly replaces itself (its size is constant).
19. The term "modernization" generally refers to improvements in the education and general status of women, and to urbanization, industrialization and the development of communications (transportation and media).
20. Period fertility is estimated at 5.8 children per woman for the years 1881-1891; it then fluctuates around this value before exceeding 6 children per woman in the early 1950s (Mari Bhat, 1989).
21. For Zachariah, 30% of the fertility decline in the period 1968-1978 resulted from the increase in age at marriage (Zachariah, 1997).
22. The contrast between north and south is widely used, but a more precise classification of Indian states distinguishes the states of the north-east, the centre, the north, etc. Reference here is to the standard distinction.
23. See Gandotra et al., 1998 for a detailed analysis of these regional differences.
24. As before, these are women aged 15-49, married or in a union.
25. The data are less recent than those for the other countries, and date from 1994.
26. The prevalence of female sterilization for contraception is slightly higher in India than in China (34.2% and 33.5% respectively) but lower than in Brazil (40%), Puerto Rico (45.5%) and the Dominican Republic (45.8%). The value for Canada is 30.6%, making it the developed country where control of fertility through female sterilization is most common.
27. For more details on the history of sterilization in India, and in particular the apparent shift away from state provision of male sterilization towards female sterilization solicited by couples, see Rajan and Véron (2006).
28. The first cases of AIDS in India were recorded in 1986. Serological tests showed that among 102 prostitutes tested in Chennai (Madras), 10 were HIV positive. However, the Indian health authorities did not take immediate steps to combat the epidemic, believing that high-risk sexual behaviour was exceptional in India (multiple partners, in particular, were uncommon).
29. The HIV prevalence rate for the adult population is 5% in Sub-Saharan Africa.
30. This is the first survey in India to include a section on HIV testing.
31. In Manipur, the HIV prevalence rate among men aged 15-54 is double that among women aged 15-49 (1.59% against 0.79%). Except in Tamil Nadu where it is lower than that for women, the male prevalence rate everywhere else is higher than the female rate, though to a lesser degree.
32. In Nepal, 83% of children aged 1-2 are fully immunized at the time of the survey, while 93% are BCG vaccinated against tuberculosis, 85% are vaccinated against measles and 89% have received the first dose of the DPT (diphtheria, pertussis, tetanus) vaccine.
33. The previous national policy on health dated from 1983.
34. In 2005, the larger size of the population aged 20-24 relative to those aged 25-29 or 30-34 is a result of high birth rates around twenty years earlier, when the relatively large cohorts of women reached childbearing age.
36. As mentioned, the war with the Soviet Union led to large-scale emigration, with many people taking refuge in Pakistan or Iran.
37. In 1960, the number of people living outside their country of birth was estimated at 75 million, or 2.5% of the world population.
38. The Indian census of 1991 counted nearly 3.4 million migrants originating from Bangladesh, with 2.6 million settled in West Bengal (Nanda, 2005).
39. The district-level map of sex ratios in the population aged 0-6 shows this phenomenon to be very circumscribed in India (Dyson et al., 2004). For an in-depth analysis, see the article by Guilmoto in this issue.
40. The role played by declining fertility in the emergence of this form of discrimination is a matter for debate (Das Gupta, 1987; Das Gupta and Bhat, 1997; Bhat and Zavier, 2003). This form of discrimination against girls has not accompanied fertility decline in Southern India.
41. The problem of food security also arises (for India, see UNFPA, 1997).
43. Data from the Asian Development Bank.