The most recent demographic data collected by India's 1992/93 National Family Health Survey revealed marked variations in fertility by state. Fertility, measured by the total fertility rate (TFR), ranged from as high as 4.82, 4.25, 4.00, 3.99 and 3.90 children per woman in the states of Uttar Pradesh, Arunachal Pradesh, Bihar, Haryana and Madhya Pradesh, respectively, to as low as the replacement, or lower, level of fertility at 1.90, 2.00 and 2.26 in Goa, Kerala and Mizoram, respectively. The national average TFR was 3.39 children per woman; the TFRs of the remaining 16 states varied between a low of 2.48 children per woman in Tamil Nadu to a high of 3.74 children per woman in the State of Meghalaya (see accompanying figure).

Various socio-demographic, cultural and economic factors may be adduced to explain these inter-state variations in fertility. The present paper is an attempt to identify the socio-economic factors, since an identification of such factors may lead to a selection of interventions amenable to policy prescriptions for narrowing inter-state variations in fertility through wider use of contraception and reduction in fertility of the states having levels of fertility above the replacement level.

____________

* The author of this paper is Adviser on Population and Development Planning and Policy, United Nations Population Fund (UNFPA) Country Support Team, Office for Central and South Asia. The views expressed in this paper are those of the author and in no way reflect the opinion of the organization with which he is currently affiliated. E-mail: cstnep@unfpa.mos.com.np

Data on births usually suffer from underenumeration, and the variations in fertility by state may also be due to the variations in underenumeration of births by state. However, in the absence of any hard evidence, there is no a priori reason to believe that the extent of underenumeration will vary from one state to another. Therefore, the variations in inter-state fertility cannot be attributed to differences in the enumeration of births between states.

The major limitation of the study is that it attempts only to account for inter-state variations in fertility behaviour at the macro level, and therefore does not take into consideration large intra-state, i.e. district, tahsil (subdistrict), ward, rural and urban, and micro or inter-individual variations in fertility. Although an analysis of intra-state and inter-individual variations in fertility would be very interesting and possibly more revealing than that of a state-level analysis, the data required to undertake this kind of analysis are not readily available. Given this constraint, the scope of the present study is confined to the analysis of inter-state variations in fertility.

The inter-state variations in fertility are examined in relation to certain aspects of female status and the survival status of children. Female status is measured in terms of a woman's access to work outside the home and education. Work status is determined using four indicators: (a) proportion of women not currently working (NW), (b) proportion of women working on family farm (WFF), (c) proportion of women working for someone else outside the home (WSE) and (d) proportion of women self-employed (SE). Educational status is measured by five indicators: (a) proportion of women literate (LR), (b) proportion of women literate but not having completed primary level education (LPE), (c) proportion of women who have completed primary level education (PE), (d) proportion of women who have completed middle school (MS) and (e) proportion of women who have completed high school (HS). The survivorship status of children is determined by two indicators: (a) infant mortality rate and (b) child mortality rate. All these data at the state level refer to the period 1992/93, and were collected by the 1992/93 National Family Health Survey of India. The infant and child mortality rates refer to average estimates for the five-year period immediately preceding the survey, i.e. 1988-1992. The infant mortality rate is the probability of dying before exact age 1, expressed per 1,000 live births. Child mortality is the probability of dying between the first and fifth birthday, expressed per 1,000 children who reach their first birthday.

Three hypotheses are posited for verification in this study, namely:

o The higher the proportion of women in a state who are working outside the home for someone else, the lower is the fertility of that state.

o The higher the proportion of women in a state who are formally educated, the lower is the fertility of that state.

o The higher the infant/child mortality levels of a state (in other words the lower the chances of survival of children), the higher is the fertility of that state.

The justifications for postulating each of these hypotheses are elaborated as follows:

Female work status and fertility behaviour

Greater female participation in the labour force has often been suggested as a means of reducing fertility. This suggestion is based on the assumption that employment outside the home provides satisfaction for women, even acting as an alternative to the rearing of children. However, female labour-force participation will not result in lower fertility per se unless there is greater incompatibility between the roles of mother and worker. Role conflict is more likely among women working for someone else outside the home than among those working on the family farm or self-employed. Self-employed women and those working on the family farm may not have to go far from home for work and therefore can combine work with rearing children. While women working outside the home for someone else may find it difficult to combine work with rearing children because of the distance of the work-place from the home, they would therefore be constrained to restrict fertility in order to continue working.

In light of the preceding discussions, it is clearly evident that among various measures of female labour force participation status, only working outside the home for someone else is likely to be negatively associated with fertility, while working on the family farm or self-employment is likely to have positive or little effect on fertility. It is therefore hypothesized that the higher the proportion of women in a state working for someone else outside the home, the lower is the fertility of that state. Alternatively, the higher the proportion of women in a state who are not currently working, or who are working on the family farm, the higher is the fertility of that state.

Female education and fertility behaviour

Education may lead indirectly to wider use of contraception and reduction in fertility by providing opportunities for personal advancement, raising aspirations for a higher standard of living, ensuring better understanding of the reproductive process, and improving access to modern and effective means of contraception as well as freeing them from traditionalism, thus enabling them to pursue modernism. The negative effect of a lack of education on fertility has been consistently borne out by both historical and contemporary data. In consideration of the mechanisms in which education may affect fertility and the universal finding of the depressing effect of education on fertility, it is also hypothesized here that the fertility level of a state will vary inversely with the level of female education of the state. And this relationship between the total fertility rate on the one hand and the level of female education on the other will hold true for all indicators of female education employed in this study.

Child survival status and fertility behaviour

It has been often hypothesized that the lower the chances of survival of children, the higher will be the level of fertility. This hypothesis is based on the assumption that, in a situation where the incidence of infant/childhood mortality is high, parents will be inclined to produce more children than necessary to ensure survival of at least a few into adulthood. In the backdrop of these arguments, it is hypothesized that the higher the infant/child mortality rate of a state (i.e. the lower the chances of survival of infants and children), the higher is the fertility of the state.

Relationships between female status and child survival status with total fertility rate

The bivariate relationship between female status and child survival status on the one hand and fertility behaviour on the other is examined using the technique of Pearsonian Correlation Coefficients (r). The value of correlation ranges between 0 and + 1; the higher the value of the coefficient, the stronger is the relationship. These coefficients are presented in table 1.

Measures of female status and child survival status	Coefficient (r)	Mean	Standard deviations
Measures of female employment status
Proportion of women not currently working outside the home (NW)	0.2015	68.5	12.85
Proportion of women working on family farm (WFF)	0.1584	12.2	9.28
Proportion of women working for someone else (WSE) outside the home	-0.4765*	16.2	8.54
Proportion of women self-employed (SE)	0.0007	3.1	6.50
Measures of female education status
Proportion of women literate (LR)	-0.6879*	53.3	316.20
Proportion of women with less than primary education (LPE)	-0.5138*	15.0	6.98
Proportion of women completing primary school (PE)	-0.6042**	12.0	4.99
Proportion of women completing middle school (MS)	-0.5952**	6.8	4.95
Proportion of women completing high school (HS)	-0.4920*	9.1	7.34
Measures of child survival status
Infant mortality rate (IMR)	0.4984*	78.5	24.77
Child mortality rate (CMR)	0.6680*	33.4	13.53

Notes: * Significant at 0.01 level.

** Significant at 0.001 level.

The coefficients in table 1 lend support to the hypothesized relationship between total fertility rate on the one hand and female status and child survival status on the other. Female education, whether measured by (a) proportion of women literate (LR), (b) proportion of women literate but not completing primary level education (LPE), (c) proportion of women completing primary level education (PE) or (d) proportion of women completing high school (HS), shows a negative relationship with fertility, i.e. the higher the level of female education in a state, the lower is the fertility of that state, and this relationship between fertility and various measures of education is found to be statistically significant.

All the measures of female employment or work status are positively associated with fertility, although they are not significant, except for the one which assesses her participation in productive activities outside the home, i.e. the proportion of women working for someone else outside the home. Employment status, measured by the proportion of women working outside the home for someone else, is found to be negatively associated with fertility, i.e. the higher the proportion of women in a state working outside the home for someone else, the lower is the fertility of that state, and this relationship is found to be statistically significant.

The data also show a positive relationship between infant/child mortality and fertility, i.e. the higher the level of infant and child mortality in a state, the higher is the fertility of the state, and this relationship is found to be strong and statistically significant, particularly the relationship between child mortality rate and total fertility rate.

We have so far examined the relationship between each of the independent variables and total fertility rate separately, i.e. at the bivariate level. In order to measure the net effect of independent variables on the dependent variable (TFR), multiple regression technique was used. While assessing the net effect of independent variables on fertility, we have selected only those variables which are strongly related to fertility and are least intercorrelated among themselves. This is to arrive at the true effect of the independent variables by minimizing the influence of multi-collinearity among the independent variables. Under the above selection criteria, independent variables chosen to represent female work, education and child survival status were the proportion of women working for someone else (WSE), proportion of women completing less than primary education (LPE) and child mortality rate (CMR), respectively. The functional form of the equation employed to measure the effects of female work status, education and child survival status on fertility is as follows:

TFRj = b0j + b1(WSEj+b2(LPE)j+b3(CMR)j+b4(SES)j+e

where TFR = average total fertility rate

j = total fertility rate of state j,

where j ranges between 1 and 25 states

b0 to b4 = coefficients

WSE = work status (i.e. the proportion of women

working for someone else)

LPE = education status (i.e. the proportion of women completing less than primary education)

CMR = child mortality rate

SES = a vector of demographic variables

e = error term

The regression results assessing the independent effect of female work, education and child survival status on total fertility rate are presented in table 2.

Variable	Regression coefficient	Standard error of B	T-value	Significance of T
Child mortality rate	0.024024	0.004806	4.999	0.0001
Proportion of women working for someone else	-0.025237	0.007384	-3.418	0.0027
Proportion of women with less than primary education	-0.028238	0.009094	-3.105	0.0056
R2 = 85.26; adjusted r2 = 82.32.

*Note: Adjusted for all the variables in the table including current use of contraception, and proportion never married.

The results of the regression analysis also confirm the earlier conclusion based on examination of the bivariate relationship in which the child mortality rate emerges as the single most important variable affecting fertility, followed by work status and education, when adjustment is made for all other variables.

The child mortality rate explains at least 18 per cent of inter-state variation in fertility over and above that which can be explained by all other variables. The total fertility rate of a state is significantly affected by its level of child mortality. The higher the child mortality rate of a state, the higher is the fertility of that state, and this relationship is statistically significant at the 0.0001 level. The implied child mortality elasticity at the sample mean was 0.237. The estimation implies that, at the sample mean, the increased child mortality rate of a state by one unit is associated with a 0.71 per cent increase in fertility of a state, and this increase is significantly different from zero.

A negative effect of women's work, i.e. the proportion of women working outside the home for someone else, on the total fertility rate is also confirmed by the data, and this relationship is statistically significant at the 0.003 level. The higher the proportion of women of a state working outside the home for someone else, the lower is the fertility of the state. For the total fertility rate, the implied work status elasticity at the point of sample mean is -0.121. The point estimate implies that, at the sample mean, increased women's participation in economic activities for someone else outside the home by 1 per cent is associated with a 0.74 per cent decline in fertility of a state, and this reduction is significantly different from zero. Women's work outside the home for someone else explains at least 9 per cent of inter-state variation in fertility over and above that which can be explained by all other variables.

Female education, measured by the proportion of women with less than primary education, also shows a strong negative association with the total fertility rate. The higher the proportion of women in a state completing less than primary education, the lower is the fertility of the state. The implied female education elasticity at the point of sample mean is 0.125 for total fertility. The point estimate implies that, at the sample mean, with a 1 per cent increase in the proportion of women in a state with less than a primary level education is associated with a 0.83 per cent decline in fertility, and this decline is significantly different from zero. Female education, i.e. the proportion of women with less than a primary level of education, explains at least 7 per cent of the inter-state variation in fertility.

The purpose of this paper has been to study the inter-state variation in fertility in relation to certain aspects of female status (education and employment) and the survival status of children (infant/child mortality). Of these three status variables, survival status of children, particularly the child mortality rate, emerges as the single most important factor explaining inter-state variations in fertility. The chances of survival of a child is strongly related to fertility: the lower the chances of survival of a child (in other words, the higher the child mortality rate), the higher is the fertility rate. However, our analysis does not show whether higher child mortality leads to higher fertility or higher fertility leads to higher child mortality.

Female labour force participation, particularly a woman's participation in activities outside the home for someone else, turns out to be the second most important variable affecting fertility. The higher the proportion of women in a state who are working outside the home for someone else, the lower is the fertility of that state. However, from our study it is not known whether women have fewer children because they like to work outside the home or they work outside the home because they have fewer children.

Female education, at less than the primary level, is the third most important variable explaining inter-state variations in fertility. The higher the proportion of women in a state formally educated, at least with a first to fourth grade education, the lower is the fertility of that state.

Data from the present study, although not presented here, also confirm a positive relationship between female education and the use of contraception. The propensity to use contraception in a state rises sharply with increases in the literacy rate of that state. And this relationship holds even when allowance is made for the effect of other variables. From these findings, it appears that the formal education of girls (for a minimum of 1-4 years) could help India go a long way towards achieving a major breakthrough in the use of contraception and in lowering fertility. However, our data do not show whether contraception and fertility among formally educated women are affected by specific knowledge, increased rationalization or the greater opportunities afforded by acquired skills, or whether education is merely a proxy measure of socio-economic status.

The study, although limited in scope, clearly points to the need for improving children's chances of survival, and also for raising women's status, guaranteeing better access to opportunities for their work outside the home and education, all of which will result in the reduction of fertility.

The average infant/child mortality rates in India, which are estimated around 79 per thousand and 33 per thousand, respectively, for the period 1988-1992, are still very high, even when compared with the corresponding rates of 23.3 per thousand and 8.4 per thousand prevailing in one of India's constituent states, Kerala.

To reduce infant/child mortality, a number of measures should be vigorously implemented: a comprehensive health and nutrition programme, especially focusing on women and children, providing for safe motherhood; universal immunization of pre-school children; and the wider availability of safe drinking water and sanitation, including provisions for supplementary food and vitamins for the children of poor parents.

To improve the educational status of women, there is a need to open more schools and to change social attitudes about female education. In a traditional society such as India's, particularly in rural India and especially among Muslims, there are some reservations about co-education. Yet there is a shortage of girls' schools. Along with the expansion of girls' schools, provision should also be made to recruit more and more female teachers in schools. Casual empiricism has shown that parents find it more reassuring to send their daughters even to co-educational schools if its faculty comprises a good proportion of female teachers. However, mere expansion of the number of girls' schools and recruitment of more female teachers will not necessarily lead to more widespread female education; social attitudes must also be changed. Indian parents generally prefer to seek higher education for sons rather than daughters, mainly because males are considered a greater economic asset to the family than females. In addition, there is a need for a persistent drive against those social customs, beliefs and traditions which belittle the value of women as compared to men and promote the value of girl children through appropriate and vigorous information, education and communication (IEC) campaigns.

Pressure groups should be formed to continuously remind the Government of its obligation to provide free and universal primary education for children by the year 2000, since India is a signatory to the Declaration on Education for All, adopted by the 1990 World Conference on Education for All. India is also a signatory to the 1992 Bali Declaration on Population and Sustainable Development and the Programme of Action adopted by the 1994 International Conference on Population and Development, both of which instruments call for the implementation of national measures such as the ones described here.

Employment opportunities should also be created for women, particularly for illiterate women in rural areas, by making provision for skills training and "soft credit" facilities. Greater employment opportunities outside the home for women would not only provide alternative satisfaction to the rearing of children but it also would provide further chances for developing new skills and techniques which may broaden their outlook and vision, and consequently lead to adoption of the small family norm and a more resolute reduction in fertility nationally.

For further information on this material please contact: loftus.unescap@un.org