Version classiqueVersion mobile

Essays on population and space in India

 | 
Christophe Z. Guilmoto
, 
Alain Vaguet

Part I. Demographic trends

2. The Geography of Fertility in India (1981-1991)1

Christophe Z. Guilmoto

Texte intégral

Fertility decline as social innovation

  • 1 This paper is part of an on-going project on fertility decline in south India, with support from th (...)
  • 2 In this paper, we distinguish crude birth rates (births per population) from more precise fertility (...)
  • 3 See Véron in this volume. On demographic transition in India and in Asia, see Srinivasan (1995), Le (...)

1The decline in fertility began more than twenty-five years ago in India and since then has had a considerable impetus. In the initial period, the effect of this decline on the birth rate was limited to offsetting a parallel decline in mortality, and the resultant growth rate remained stable.2 Since the late 1980s, the birth rate has been diminishing more rapidly than the rate of mortality and the theoretical natural growth of the Indian population, that is, potential growth in the absence of migration, has begun to significantly decrease, to the extent of being reduced to zero in some areas of the country. This is without doubt a matter of an historical stage in the evolution of India, parallel to the more general movement to be observed in Asia.3

  • 4 On fertility, see the chapters by Willis and McNicoll in Kiessling and Landberg (1994) For the comp (...)

2From the demographic point of view, fertility is the consequence of different parameters of a socio-demographic nature, among which are marital life, recourse to contraception and abortion, couple fecundity and intra-uterine mortality. Only the first two factors (or “intermediate variables”) in this list play a significant role regarding the fertility level, and contraceptive practice is, moreover, the indicator most closely associated with fertility in countries where the latter has already decreased, In India, the government began in the 1970s to initiate important policies pertaining to birth control, favouring the availability of contraceptive methods (female and male sterilization, IUD, condom, etc.), as well as the propagation of a new ideology of the family. To understand the choices regarding reproductive behaviour, it is, however, necessary to take into consideration a complex set of factors, ranging from the economic considerations of households regarding the advantages of few children (demand factors), by way of the system of norms and ideals relating to fertility, to the actual availability of contraceptive methods (supply factors).4

3Adopting a sociological perspective, the decline in Indian fertility is seen to be among the innovation phenomena that are transforming social customs. This takes place according to diffusion mechanisms, the process of which is undergoing a number of temporal and spatial irregularities. The almost irreversible character of the declining evolution of fertility imparts to it a progressive impetus, both geographical and social, within a definite space characterized by the occurrence of diffusion. One will thus be able to identify the original core areas of diffusion, then the line of progress which innovation follows until the areas have been exhausted. The fertility level will thus be equivalent to the proportion of those who are observed to adopt innovations, with an initial pioneer phase, followed by a longer phase during which the area of contact broadens, before returning to a stable state subsequent to saturation.

  • 5 For very distinct analyses of the role played by diffusion processes in fertility decline, see Clel (...)
  • 6 About diffusion mechanisms, see for example Rogers (1995) and Saint-Julien (1992).

4The analysis in terms of diffusion complements the more common interpretations of the decline in fertility, which emphasize structural factors characteristic of social and economic changes.5 One knows, in fact, the influence of “modernization” on fertility, corresponding, for instance, to urbanization, industrialization and, above all, to progress in education. However, it today appears more clearly that, independently of structural changes on a macro level, the decline in fertility typically proceeds by contagion, as the spatial diffusion of new attitudes and behaviour, for which any form of proximity, spatial or social, is an important factor of impetus. Conversely, social, cultural or geographical isolation, that is, the low density of exchanges outside the reference group, constitutes one of the first barriers to the spreading of innovations, imposing on them a considerable number of boundaries or stages to be crossed.6

5The strong heterogeneity of the Indian environment, notably in its geographic and social dimensions, suggests as a consequence that the introduction of new ideas or behaviour cannot take place in a uniform manner according to milieu and that, subsequently, the diffusion of innovations will initially generate new forms of social differentiation. This is precisely the lesson to be derived from a geographical approach to fertility decline in India. Reproductive behaviour, in changing, creates new inter-regional disparities which represent as much the inherited historical and cultural diversity, as the disequilibria to come. The object of this article is precisely to explore this new diversity on the basis of various maps and statistics pertaining to fertility, while considering the lessons derived therefrom for Indian cultural geography. The progressive and irregular penetration of Malthusian behaviour in the country reflects certain dimensions of its spatial organization ; these in turn refer to family models (kinship structures, the position of women, etc.) which vary according to each regional culture.

The spatial dimensions of fertility

  • 7 While differences between fertility estimates between Indian states used to be minimal between the (...)

6A study of disparities concerning the demographic pattern in India has traditionally led to the opposition of pioneer states, of which Kerala constitutes the spearhead, to the states which lag behind, where the fertility level appears to have been only slightly affected by the extensive birth control campaigns which were begun in the 1960s. The diffusion of new behaviour doubtlessly introduced a new heterogeneity to the demographic landscape of India, for the differences in fertility among regions, although not strictly speaking negligible prior to the beginning of the demographic transition, were nevertheless of lesser magnitude. The few statistics which enable us to describe the former regional disparities in fertility indicate, in effect, that the variations between regions, in particular linked with early female age at marriage, was minor before Independence, especially when compared with the new situations in the 1980s, which our cartography will illustrate.7

  • 8 See Anderson (1974). This is one of the only studies to use district-level census statistics. It ma (...)

7The most precise cartography available for the period following Independence remains ambiguous, for it relies on distributions based on age groups with unstable behaviour.8 It will be noted, however, that the regions with low fertility in 1951 and 1961 appear to be distributed, according to analyses by Anderson, along the Bay of Bengal, from the Thanjavur region to coastal Andhra and Orissa, with another pocket located in the Garhwal region (Uttarakhand). Conversely, the areas with the highest recorded fertility values are more scattered, covering the Punjab, the valley of the Brahmaputra in Assam and a few isolated areas of Bihar and Madhya Pradesh. This map of pre-transitional fertility is difficult to interpret because the data are both inconsistent and incomplete. The spatial patterning of fertility is very fragmented because the distribution of values is segmented around different characteristic regions, such as the Punjab, Assam or Tami Nadu. Regional demographic identity seems to take precedence over global spatial principle. Nevertheless, some geographical orientations in fertility appear to foresee features illustrated by recent maps, the north-south opposition, for instance.

  • 9 In 1991, India comprised 466 districts (average population of 18 lakhs) and around 3,500 taluks (av (...)
  • 10 See Bhat (1998) ; Malhotra et al. (1995) ; Murthi et al. (1995).
  • 11 See Sopher (1980) ; Karve (1968).

8Recent statistics show very strong differences, for fertility doubles as one moves from the “advanced” states such as Kerala or Tamil Nadu, where fertility is today at about the replacement level (two children per woman), to the large Hindi-speaking states such as Uttar Pradesh, Bihar, Rajasthan and Madhya Pradesh, where fertility is significantly higher. We intend to examine here these differences on a geographical scale which is smaller than the states, the size of which (fifteen of them have populations exceeding ten million inhabitants) could conceal a high degree of heterogeneity.9 Various statistical analyses of fertility differentials carried out on the basis of figures from 1981 or 1991 have demonstrated the major role played by the status of women, above all expressed by literacy rate and activity, in the determination of local fertility levels.10 It should be noted that despite the great variety of indicators and variables tested, the measurable economic differences between Indian regions appears to occupy only a very secondary place in the establishment of fertility models for the districts of the country. These observations tend toward the surmise of Dyson and Moore (1983), who were the first to relate fertility differentials in India to its social and cultural structure. These authors, following studies by Sopher in geography and by Karve in ethnology, emphasized the crucial role of the virtual north-south boundary separating Sanskritized and Islamized north India from Dravidian south India, strong in its specific cultural traditions.11 We shall return later to the interest as well as the limitations represented by this division of India. The fact remains that these recent statistical models, with an abundance of social, cultural and economic variables, have not succeeded in entirely freeing themselves from “residual” geographic dimensions, in particular from the north-south divide.

Fertility differentials on the scale of districts

  • 12 See International Institute for Population Sciences (1995). For a new regional analysis of NFHS, se (...)

9Before proceeding with a cartographic examination of Indian demography, we shall consider from a statistical point of view the corpus of available data pertaining to fertility, thereby underscoring the heterogeneity that characterizes it. As a result of inadequacies in the civil registration (incompleteness), the most reliable data relating to Indian fertility are derived from sample surveys. It is a question in this case of the annual evaluations of the Sample Registration System (SRS), which provide basic indicators regarding mortality and birth rate, and from a national campaign of detailed demographic and health surveys conducted in 1992-93 (National Family Health Survey). These sources provide relatively sound estimates, both for tendencies since the 1970s and the social determinants of demographic behaviour.12 However, for our purpose they are very imprecise geographically, for the statistics from them are aggregated on the level of the states, which have, as previously mentioned, populations much too large to enable of cartographic study.

  • 13 See Mishra et al (1995). A new set of estimates is now available for 1991 : District Level Estimate (...)
  • 14 Bhat’s method for estimating the total fertility rates consists in a backward projection of the pop (...)
  • 15 This procedure enables to represent simultaneously levels (averages) and variation (variance) level (...)

10On a smaller geographical scale and because of the various inadequacies already noted in the civil registration, only census data can, on the basis of their exhaustiveness, compensate for this statistical deficiency. However, the census is not conceived to measure fertility and, as a consequence, one must utilize indirect measurements derived in general from the proportion of children of an age below seven years, or from questions pertaining to births during the year preceding the census.13 Based on the data pertaining to children below seven years in 1991, Bhat has recently proposed a series of fertility estimates for 326 districts in the largest states of India.14 We shall use these figures to give a general idea of the geographical differentials. To do this, we have chosen to represent the data pertaining to the fertility of the fourteen largest states in the form of a boxplot. A brief explanation is required to understand this type of graph, which is intended to comparatively summarize the distribution of a characteristic, in this instance the fertility level per district assessed in 1991.15 The different states appear on the abscissa and are here classed, for the sake of simplicity, by increasing level of fertility. Fertility indices are measured on the ordinates in terms of the number of children per woman ; for each state, the whole comprised by the districts was divided into four quartiles : the first quartile consists of the districts having the lowest values, and thus successively until the fourth and last quartile which consists of the districts with the highest values. Each “box” appearing in Figure 2.1 represents the fertility of half the districts, that is, the second and third quartiles. The fertility value in the “median” district appears in the centre of each box. The box, therefore, represents the central tendency and the intra-regional variance (among the districts of the same state).

Figure 2.1 : Distribution of district fertility estimates for longer states, 1991.

Figure 2.1 : Distribution of district fertility estimates for longer states, 1991.
  • 16 These results are based respectively on an ANOVA performed by state and on an ordinary least-square (...)
  • 17 The heterogeneity is a function of the total number of units. On Figure 2.1. the width of each box (...)

11The graph (Figure 2.1) confirms, first, the contrasting character of fertility and the diversity of situations, from the two Malthusian states of the southern tip of India, to the four most fertile states in the north. These states appear, moreover, very much out of line with the rest of India. An analysis of the variance in fertility furthermore indicates that the division into states accounts for 70.6 per cent of the variation between districts ; this result signifies that the regional patterning has a statistical weight for fertility which is comparable or superior to that of literacy, the socio-economic variable most closely correlated with the number of children per woman.16 Nevertheless, the graph also enables of an evaluation of demographic diversity even within these large regional units. One will observe, for example, that the values can be very narrow in some states, such as Tamil Nadu, the Punjab, Bihar or even Uttar Pradesh, which has the largest number of districts in India.17 This situation suggests a higher degree of homogeneity within these administrative units. Conversely, states such as Karnataka, Maharashtra, Rajasthan or Madhya Pradesh appear to be particularly heterogeneous in terms of reproductive behaviour. These disparities among sub-regions of the same state, when they are also pronounced, are indicative of particularisms which characterize local societies on a smaller scale.

12These initial observations pertaining to the central dispersion are sustained by the examination of the “extreme values” represented in the graphs by the two horizontal bars (also called “whiskers”) located below and above the box. These bars are an extension of the central box (one and one half times the central dispersion) and make it possible to also identify a few isolated extreme values which are beyond their range. Thus, whereas the median value for fertility in Bihar or Uttar Pradesh is above five children per woman and the variance is rather limited, several districts (often urban) are to be distinguished by atypical fertility values of approximately 3.5 children per woman.

13This examination, based on a sample encompassing 350 districts, thus moderates the stress placed alone on the level of the state by geographical analyses. Although Figure 2.1 effectively shows that the different Indian states are in much differentiated situations as concerns fertility transition, numerous intra-regional differences remain. Thus, to again take up the states in all-contrasting points, one will observe that in some districts of Uttar Pradesh and Bihar, fertility is comparable to that in a district of Kerala. This analysis could be pursued on a smaller scale, comparing the situation of the taluks of the same state. We have carried out such an analysis for the rural areas of Tamil Nadu, which is one of the least heterogeneous states with respect to fertility, as indicated by our preceding analysis. The dispersion of values obtained for the child/woman ratios remains quite high, varying twofold from taluk to taluk. We are thus confronted with an extremely contrasting demographic landscape, on the regional as well as local level, which in particular denotes the yet incomplete character of the penetration of new reproductive norms in the states, small regions, towns, urban quarters or localities. The following cartography makes it possible to discover certain principles of the spatial organization of fertility in India.

The spatial dimensions of fertility transition

  • 18 For the maps, 1981 values have been adjusted for new 1991 districts. When district level estimates (...)

14The two maps presented here (Figures 2.3 and 2.4) employ fertility estimates by Bhat (1998) ; we have held to one homogeneous source so as to limit the discrepancies which could result from different methods of evaluation. The main weakness of these figures lies in the nature of the evaluation of infant mortality. It will be noted that the figures, derived from decennial censuses, refer to fertility during the six preceding years. The classification of fertility levels remains the same from one map to the next, but some values are entirely lacking for areas where the surveys could not be carried out (Assam in 1981, Jammu and Kashmir in 1991).18

15The maps present a coherent and regular picture of the distribution of Indian fertility, with regional variations, which are sufficiently pronounced to enable of a detailed geographical interpretation. Since the late 1970s (Figure 2.3), average fertility has fallen below 3.5 children per woman in several areas. The first of these Malthusian pockets is evidently centred around Kerala, or more precisely around the former regions of Travancore and Cochin, which recorded the lowest values s in the country. Other adjoining districts also reached this low level of fertility, namely the historical centre of Tamil country (Coimbatore, Madurai, Thanjavur) and the nearly continuous coastal strip of the Konkan extending from northern Kerala, by way of Goa, to Mumbai (Bombay). The adjacent regions, forming a large crown bordering the coast from Gujarat to Andhra Pradesh and southern Bengal, clearly seem to have entered this declining evolution, but less sharply because the average number of children remains above 3.5. Other areas of decline are also perceptible, notably in the Punjab, Himachal Pradesh and Manipur. With the exception of Mumbai, some urban centres appear to be strongly affected by the decline, such as Chandigarh, Pune, Chennai (Madras) and Calcutta, often with an effect on their rural hinterlands.

16This profile has become remarkably pronounced in a period of ten years (Figure 2.4). Fertility in 1991, at less then 2.5 children per woman, henceforth entered the final phase of its transition in the southern promontory of the subcontinent. The decrease in fertility was particularly spectacular in Kerala and Tamil Nadu, even if some small regions, for instance Malabar, remained slightly behind. The same holds for the districts of Goa, Bangalore and the Karnataka coast. Elsewhere in India, equally low fertility levels are scarcely to be found, except in some metropolitan agglomerations such as Chandigarh, Mumbai and Calcutta. In these latter areas or cities, fertility decline, following the example of the experience in European or Asian countries, must not be long in reaching a lower limit of about 1.5 children per woman. As indicated by the results of the NFHS survey in Kerala, Goa and Tamil Nadu, the aggregate of couples there have recourse to contraception in order to space and limit births, while the ideal of fertility has drawn very close to the reproduction level, that is, two children per couple. Although the very young age structure still lends these populations a strong demographic impetus, growth will soon be very low, indeed nil, in numerous districts.

17The general downturn in fertility in the regions identified ten years earlier also increased, to the extent of covering nearly all of coastal and southern India with fertility levels of less than 3.5 children. In the interior of coastal states, such as Maharashtra and West Bengal, districts with highly variable fertility levels nevertheless coexist. The decline also continued in the pocket of the north-west, centred on Chandigarh and rural pockets in the Punjab, without having registered as great an impact in the Hindi-speaking areas of Haryana and Uttar Pradesh. The very rapid urbanization of Delhi and its region appears, moreover, to have only a modest effect on demographic behaviour. It is, however, true that the capital is next to the zone which shows the greatest resistance to the diffusion of new attitudes and reproductive practices, about which a few words should be said.

Figure 2.2 : Region and cities mentioned in the text

Figure 2.2 : Region and cities mentioned in the text
  • 19 viz. Bihar, Madhya Pradesh, Rajasthan and Uttar Pradesh. This designation has been introduced by As (...)

18This region, which in a way constitutes the core of the traditional demographic system, characterized above all by a high fertility and a vigorous patriarchy, has grown smaller in the interval of two censuses, but conserves a strong spatial cohesion within the so-called Bimaru zone, which is comprised of the four large states of north India.19 It includes the western fringe of Uttar Pradesh, a rather prosperous agricultural region bathed by the Ganga and the Yamuna (Doab), then extends south toward Rajasthan and above all toward Madhya Pradesh, of which it forms notably the northem border (Chambal valley, Bundelkhand and Baghelkhand), stretching toward Bihar. This region occupies a central position in the Hindi-speaking zone of the Ganga basin, which includes Haryana and Uttar Praand as well as by extension Rajasthan, Madhya Pradesh and Bihar, the regional or local dialects of which (Rajasthani, Bhojpuri, Urdu, etc.) are very close to Hindi. It is particularly characterized by its pronounced economic underdevelopment, the numerical importance of Brahmins and the marginalization of women, both in terms of education and in terms of employment. However, on other planes, it is far from being completely homogeneous, in particular as concerns social composition (number of Muslims, tribals), density and urbanization or landscape (plains, hills, forests). Only a few isolated districts, notably in the very remote mountainous state of Arunachal Pradesh, record today fertility levels as high as in the Bimaru core area, above five children per woman, levels which have subsequently only marginally evolved since fertility decline began.

A spatial interpretation of fertility

19The dispersion of fertility in India, as interpreted by means of these two maps, presents a sufficiently manifest geographical regularity to provide now a few main principles. Thus, if one does not take into consideration the demographic geography of the states in the north-east, which is, moreover, not well known, the strong compactness of demographic change reflects an image of Indian fertility as successive crowns centred around the Gangetic region, which has the highest fertility. The distance from this centre is, moreover, the first key to interpreting the diffusion, originating on the periphery, of the decline in fertility. We thus propose, on the basis of these maps, to substitute for the classic north-south dichotomy, popularized in particular by Dyson and Moore, a more complex spatial structuration, defined by an almost concentric distribution of fertility around a central locus at the junction of Hindi-language states. Although the most distant regions, in this case the southern promontory of India, are indeed those in which the reduction of fertility has been the greatest, the principle of gradation is also applicable to the nearest peripheral areas, such as the mountainous regions of Himachal Pradesh and Uttar Pradesh (Garhwal). The movement in the last twenty years of the “front line”, which delineates the belt of high fertility areas, serves to accentuate this crown-shaped organization of Indian space, but also the progressive isolation of the regions most favouring rising birth rates which had earlier formed a much more extensive space, including even border countries such as Nepal, Pakistan and Bangladesh (Aggarwal, 1994 : 368)

Figure 2.3 : Fertility estimates for 1981 due to Mari Bhat

Figure 2.3 : Fertility estimates for 1981 due to Mari Bhat
  • 20 On the role of the Indian littoral, see Sopher’s interpretation in the Geographical Patterning ot C (...)

20A second key to the interpretation of our maps relates to what we shall call the “littoral effect”, that is to say, the particularly rapid rate which characterizes the transition in the coastal areas, perhaps with the exclusion of Orissa. This phenomenon can of course be interpreted in terms of distance from the seat of high fertility on the continent, but also in geo-historical terms. The coastal regions are, in fact, those where exchanges with the rest of the world, notably because of colonization, were the most intense. Their population also often distinguishes itself from the inland areas (earlier conversions to Christianity and Islamization, castes of fisherfolk, etc). and progress in education has been much more rapid there.20 It could seem difficult to distinguish the littoral effect from the centrifugal gradient mentioned above, insofar as these two principles of spatial organization appear to describe the same degree of proximity to or distance from the core. However, it will be noted that unlike the classical example of diffusion spending outwards from innovation modes, the change has a peripheral origin.

21A third rule, independent of the two first principles, follows from the focal role which falls to a large number of metropolises in the diffusion of new reproductive behaviour, in particular agglomerations with a high industrial concentration. We have already mentioned the effect pertaining to Calcutta, Chennai and Chandigarh. But a closer reading also brings out the influence of Bhubaneshwar in Orissa, Bangalore in Karnataka, Nagpur in Vidarbha (Maharashtra), Rourkela and the other industrial towns of Chota Nagpur, Patna in Bihar, Lucknow and Kanpur in Uttar Pradesh, etc. If it is once again a question of a gradient effect, this principle obviously also holds on a smaller scale, for the influence of towns is rapidly diluted as soon as one moves away from them. This “urban effect” would also doubtlessly appear more clearly on the micro-regional scale. Moreover, it remains to explain why the effect of certain metropolises sometimes appears to be imperceptible or very weak in the interior of regions with a high fertility (around Delhi or Jaipur), or regions with a low fertility (around Kochi or Coimbatore).

  • 21 Various analyses of fertility differentials for 1981 and 1991 are compared in Guilmoto and Rajan (1 (...)

22Alongside these general guiding principles, certain regional particularisms are still perceptible. It is thus no longer a question of a general trends, but of a patterning in more homogeneous sub-regions which are clearly distinguished from neighbouring areas. The pioneering and atypical role of the Punjab has already been noted in the fertility decline in north India. Similarly, an apparently rapid decrease was recorded in Telengana (Andhra Pradesh), Goa and Saurashtra (Gujarat). Conversely, pockets of high fertility subsist in more extensive areas where changes were, on the contrary, very rapid. This is notably the case over an area of the western part of the Deccan plateau, along a band extending northward to Maharashtra (Marathawada) as far as Karnataka (Hyderabad Karnatak), and closely corresponding to the implantation of the Muslim sultanates until the 18th century. Similarly, the peripheral tribal areas of Madhya Pradesh (Chhatisgarfi, Gondwana, and Bhil country) report a singularly high fertility rate. One may think to discern the effect of sociological profiles of a type favouring high birth rates (presence of tribals or Muslims), or of a Malthusian type (Christian and Sikh presence) in the determination of these regional disparities. Nevertheless, the effects of these factors on fertility are ambiguous, as indicated by statistical analyses based on less impressionist examinations of the Indian demographic landscape.21

  • 22 More detail in Malhotra et al. (1995) or Murti et al. (1995). For a discussion on spatial statistic (...)

23This observation will call to mind that, as concerns the limitations of our interpretation based solely on cartographic representations, compared with the contribution of statistical models, the spatial configuration of fertility does not of itself enable one to confirm the connection between demographic behaviour and the other social, cultural and economic dimensions of the Indian landscape. On the other hand, some specifically spatial dimensions of demographic change, which we have just considered, scarcely emerge from statistical analyses. One must, in fact, be able to evaluate the specific effect of the spatial configuration on the relationships between variables. The only tools presently available for this undertaking concern the measure of spatial autocorrelation. In the case of the analysis of the regional determinants of fertility, the models that have integrated this spatial dimension have clearly underscored the intensity of this spatial autocorrelation.22 The cartographic examination which has been conducted confirms this, for the very high level of covariations between border districts is evident, and this is beyond the strong homogeneity in the interior of the states evaluated by means of Figure 2.1.

  • 23 The higher fertility levels observed in some districts of Kerala (such as Mallapuram) may be relate (...)

24This spatial autocorrelation is, however, manifested in a more complex manner than by the effect of spatial proximity alone. In effect, the sociocultural boundaries (population, linguistic group, historic substratum) create very abrupt points of continuity, as, for example, going from coastal Andhra (Godavari delta) to the Bastar region (Madhya Pradesh) ; or, without leaving the state of Gujarat, going from the town of Vadodara (Baroda) to Panch Mahals district. These stages indicate the presence of invisible boundaries, as for instance that of tribal India in the examples which we have mentioned. Conversely, numerous identified areas, which consequently correspond to demographic territories having a district identity, span several states, sometimes crossing over well-established social or physical borders. The central region of high fertility is a prime illustration of this. One may take as an additional example the most advanced centre of fertility decline in India, on either side of the southern Ghats. The homogeneity of this area respective of fertility, with henceforth less than two children per woman, would lead one to overlook the linguistic heterogeneity (Tamil-Malayalam) and the natural border of the Ghats.23 As regards the neighbouring regions, such as the rest of Tamil Nadu or southern Karnataka, the discontinuity of fertility values is pronounced.

25The cartographic study therefore makes it possible, first, to identify certain less known boundaries, which divide Indian space in the interior. This pattern, which fertility makes perceptible, doubtlessly does not correspond to that which other phenomena could bring to light. But insofar as the decline in fertility is indicative of the penetration of a new system of norms, and more particularly of new parameters in the equation which links individual and familial reproductive behaviour with social and economic contexts, the unequal penetration in India of Malthusian behaviour illustrate the new social differentiation which modernity could introduce, above and beyond the better known differences in terms of social classes or of type of residence (urban/rural). It would subsequently be appropriate —but this project exceeds the descriptive intention of this article to relate this map to that of other dimensions of social change (electoral behaviour, penetration of wage-earning and market economy, etc.).

26In the second place, the light shed by cartography also provides a diachronic dimension to the study of the propagation of changes. One will easily observe, in fact, that the spatial disparities shown on the map of 1981 indicate the form of changes noted ten years later, in such a way that the map of 1991 enables one to similarly foretell the changes in fertility during the current decade. Although it is certainly hazardous to predict the effective magnitude of decline, the regional demographic panorama in the year 2000 is quite foreseeable : fertility will doubtlessly come down throughout south India (with the exclusion of the formerly Muslim Karnataka) and on the coastal perimeter of Gujarat and Maharashtra to the values typical for industrialized countries (two or less children per woman). The centre of fertility decline in the Punjab will extend to Himachal Pradesh, northern Rajasthan, Haryana and Delhi, fertility reaching then values lying between two and three children per woman. Selectively, as around Chandigarh, fertility rates could even decline more significantly. Similarly, the decline in fertility would be intensified in the region of Nagpur, Calcutta and some states in the north-east (Manipur, Mizoram, Tripura), and the number of children per woman would approach the replacement level of generations (approximately 2.2 children per woman). Elsewhere in India, the effect of the decline will be less spectacular, for levels will probably remain above three children per woman. It nevertheless remains difficult to foresee the magnitude of decline in areas more resistant to the diffusion of new demographic behaviour. In addition, the population density of the Gangetic basin lends a considerable demographic weight to this question for the entire country.

  • 24 For international comparison, see Jejeebhoy (1995).

27From a more theoretical point of view, the geographic approach brings with it a few additional interpretative points important in the classic analysis of fertility decline in India. The latter has brought to the fore diverse social and cultural dimensions, notably the status of women in society and in production. Literacy and the activity of women are the variables most closely associated, from a statistical point of view, with a lower fertility. Expressed in more analytical terms, these dimensions of the status of women represent the capacity of women to escape the traditional definition of their role as mothers and wives through a higher degree of social autonomy and a better integration in the labour market.24 Fertility would therefore be principally a response of households in the interior to this progressive emancipation. However, the strong geographic coherence, which emerges from the cartographic examination the independent role, played by the spatial mechanisms characteristic of the process of diffusion. The transition of fertility has, in fact, followed a course typical of the diffusion of innovations along the channels of social change, relying thus very heavily on social and cultural contiguities in order to spread. Government intervention, whether a question of diffusion of contraceptive products or, in a more nonmaterial manner, that of a new system of norms giving greater importance to the quality of children (health, education) over their quantity, has consequently received responses which are highly variable from region to region : the geographical profile which results thereof disregards the fact that it would only be a matter of unequal expertise (and assiduity) of the different regional governments in the implementation of the family planning policies decreed in New Delhi.

Figure 2.4 : Fertility estimates for 1991 due to Mari Bhat

Figure 2.4 : Fertility estimates for 1991 due to Mari Bhat

Conclusion

28This geographical presentation of the unequal progress in the decline of fertility in the interior space of the country reinforces the interest in the study of the mechanisms of diffusion at work in the redrawing of the demographic map of India. However, some questions remain in abeyance to which our analysis cannot respond. Notably, one does not know the precise locus of the demographic innovation because numerous scales of interpretation exist regarding changes in reproductive behaviour : women, households, communities, or social groups. Our level of analysis (that of the district) obviously remains too global to test precise hypotheses, for one could easily be victim of a classic illusion of scale (ecological fallacy). In the description of diffusion, one also remains confined by an approach which is too global to be able to identify the relays and channels characteristic of the diffusion of new familial representations and contraceptive techniques. As mentioned earlier, the first analysis on a smaller scale, on the level of taluks, or even of panchayats which constitute the former, indicates that the degree of heterogeneity of behaviour remains quite pronounced on the local level. With the help of a more detailed micro-regional cartography, one should be able to bring to light the respective roles of institutional intervention (family care and planning centre) or of communication infrastructures (road network, cinemas, etc.), in the dynamics of the demographic landscape.

29The cartographic scale employed for this article nevertheless makes it possible to shed light on a few fundamental features of a new spatial dynamics in India which call into question classic models of cultural geography. The segmentation in autonomous enclaves, legacy of the historical morphology of Indian space, seems to subside gradually. One has certainly isolated the specific identity of a few regional pockets and the effect of persistence on fertility, as for instance in the former sultanates in the Deccan, but this mosaic” division plays a minor role. The different forms of modernization which can be selectively supported by regional identities (political, social, cultural, etc.), as the example of Kerala calls to mind, impose new principles of organization, and the mechanisms of exchange at the centre of the diffusion process have precisely substituted more ample and constantly shifting boundaries for the strong historical disparities. In addition, Indian space has henceforth changed from a mosaic to a crown-shaped structure, which seems to be principally oriented around the hard core of traditional India, where fertility has scarcely declined.

30In such a structure, the least customary feature is indubitably the peripheral (exogenous ? ) origin of innovation and the centripetal progression of birth control, which starts from the coastal fringes of Indian space and moves toward the centre of the Gangetic valley. Contrary to the centrifugal action of propagation which gains ground starting from the origin, the process of diffusion from the periphery toward a centre of resistance is more similar to a gradual encircling of a central region which appears to be besieged by modernity. Let us note in conclusion that the growing heterogeneity of fertility behaviour which we have brought out in the course of this article conceals, in fact, a future transregional homogenization of demographic behaviour on the pan-Indian scale, that is to say, a process of progressive convergence toward a dominant Malthusian model as in other Asian countries (Sri Lanka, Thailand, South Korea), effacing regional discontinuities to the advantage of a global geographical readjustment. Fertility doubtlessly illustrates today how Indian space integrates this homogenization and how the resistance to these social upheavals is organized.

Notes

1 This paper is part of an on-going project on fertility decline in south India, with support from the French Institute, the Research Institute for Development (IRD) and the Wellcome Trust.

2 In this paper, we distinguish crude birth rates (births per population) from more precise fertility rates computed as average number of children per woman. The measurement of birth rates is affected by other factors such as the age- and sex-distribution of the population.

3 See Véron in this volume. On demographic transition in India and in Asia, see Srinivasan (1995), Leete (1993), Martine (1998). Fertility decline may have started earlier in some areas. See Das Gupta (1995).

4 On fertility, see the chapters by Willis and McNicoll in Kiessling and Landberg (1994) For the comparison of different settings of fertility decline, see : Egerö and Hammarskjdöld, (1994) and Martine et al. (1998).

5 For very distinct analyses of the role played by diffusion processes in fertility decline, see Cleland and Wilson (1987) and Friedlander et al. (1991).

6 About diffusion mechanisms, see for example Rogers (1995) and Saint-Julien (1992).

7 While differences between fertility estimates between Indian states used to be minimal between the 1960s (a ratio of 1 to 1.2-1.4), these are now considerable (1 to 3). See Visaria and Visaria (1982), p. 509. Srinivasan (1995) pp. 67-94.

8 See Anderson (1974). This is one of the only studies to use district-level census statistics. It may be noted that the probable fertility increase that took place in 1951-61 does confuse the picture.

9 In 1991, India comprised 466 districts (average population of 18 lakhs) and around 3,500 taluks (average population of 2,4 lakhs). Demographic data below state level are infrequent.

10 See Bhat (1998) ; Malhotra et al. (1995) ; Murthi et al. (1995).

11 See Sopher (1980) ; Karve (1968).

12 See International Institute for Population Sciences (1995). For a new regional analysis of NFHS, see the estimates for 76 natural regions in Bhat and Zavier (1999).

13 See Mishra et al (1995). A new set of estimates is now available for 1991 : District Level Estimates of Fertility and Child Mortality for 1991 (1997).

14 Bhat’s method for estimating the total fertility rates consists in a backward projection of the population of children below seven for 1981 and 1991. His method takes into account variations in district-level mortality. See Bhat (1998).

15 This procedure enables to represent simultaneously levels (averages) and variation (variance) levels. See Tukey (1977) ; Bhrolcháin and Toulemon (1996).

16 These results are based respectively on an ANOVA performed by state and on an ordinary least-square regression for 361 districts.

17 The heterogeneity is a function of the total number of units. On Figure 2.1. the width of each box has been made proportional to the (square root of) number of observations.

18 For the maps, 1981 values have been adjusted for new 1991 districts. When district level estimates were not available (as in north-east India), I have used state averages. Islands are not shown on the map. Other maps for 1981 or 1991 may be found in Mishra (1995) and Malhotra (1995).

19 viz. Bihar, Madhya Pradesh, Rajasthan and Uttar Pradesh. This designation has been introduced by Ashish Bose. For a demographic analysis of this macro-region, see Sathia and Jejeebhoy (1991). This study unfortunately does not examine internal demographic variation within these large states.

20 On the role of the Indian littoral, see Sopher’s interpretation in the Geographical Patterning ot Culture in India”, in Sopher (1980).

21 Various analyses of fertility differentials for 1981 and 1991 are compared in Guilmoto and Rajan (1998).

22 More detail in Malhotra et al. (1995) or Murti et al. (1995). For a discussion on spatial statistics, see Charre (1995) ; Hainning (1990) ; Bailey and Gatrell (1995).

23 The higher fertility levels observed in some districts of Kerala (such as Mallapuram) may be related to the higher proportion of Muslims in these areas.

24 For international comparison, see Jejeebhoy (1995).

Table des illustrations

Titre Figure 2.1 : Distribution of district fertility estimates for longer states, 1991.
URL http://books.openedition.org/ifp/docannexe/image/9803/img-1.jpg
Fichier image/jpeg, 38k
Titre Figure 2.2 : Region and cities mentioned in the text
URL http://books.openedition.org/ifp/docannexe/image/9803/img-2.jpg
Fichier image/jpeg, 96k
Titre Figure 2.3 : Fertility estimates for 1981 due to Mari Bhat
URL http://books.openedition.org/ifp/docannexe/image/9803/img-3.jpg
Fichier image/jpeg, 104k
Titre Figure 2.4 : Fertility estimates for 1991 due to Mari Bhat
URL http://books.openedition.org/ifp/docannexe/image/9803/img-4.jpg
Fichier image/jpeg, 93k

Auteur

Institut français de Pondichéry and Institut de Recherche pour le Développement, Pondicherry, India.

© Institut Français de Pondichéry, 2000

Conditions d’utilisation : http://www.openedition.org/6540

Cette publication numérique est issue d’un traitement automatique par reconnaissance optique de caractères.

Acheter

Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search