Impact of Women’s Education on Fertility: Evidences from NFHS-II Data

 

 

 

 

By

Rajnikant. M. Patel

Devamoni Dey

 

 

 

 

 

 

Population Research Centre

Department of Statistics

Faculty of Science

M. S. University of Baroda

 

Paper prepared for presentation at the 25th Annual conference of Indian Association for the Study of Population to be held at IIPS, Mumbai, 11th to 13th Feb., 2002


Impact of Women’s Education on Fertility: Evidences from NFHS-II Data

 

INTRODUCTION:

Among various dimensions of women’s status education deserves special attention, since, it largely conditions the equality of women’s lives and is highly susceptible to improvement through policy intervention.  Although, for women’s education should be promoted primarily on human rights and social justice grounds, it is widely recognized as one of the most promising catalyst of sustained fertility decline.  Knowledge on the casual linkages between education and fertility can hence serve as a valuable policy instrument in the contest of development planning.  Over the past decades, the need to promote women’s education has been repeatedly advocated in both international and national political forums. The United Nations decade for women: Equality, development and peace (1976-1985) brought to the forefront of global agenda, the necessity to achieve full integration of women in society on an equal basis with man and increased worldwide awareness on women’s educational needs. 

            Although education has been unanimously endorsed as fundamental right and as an explicit developmental objective, women’s access to educational resources remains inadequate   in large part of the developing world (Kelly and Elliott, 1982). In India, a long historical neglect of women’s educational need has left a legacy of very high illiterate rates, especially among the older, poor and rural women.  During the past decades the Government as committed itself to expanding women’s educational facility, raising girls enrolment ratios and guaranteeing equal access to schooling. As a result, although institutional and legal barriers to women’s access to education have been increasingly removed, deep rooted cultural beliefs and social habits that sustained gender inequality have a prolonged inertia.  Despite all obstacles, education continues to inspire big hopes as the most promising strategy to enhance women’s status.

            Besides its role as a development strategy, women’s education has long been recognized as a crucial factor in reproductive behaviours.  The World Population Plan of Action, adopted at the United Nations Plan of Action, adopted at the United Nations World population Conferences held in Bucharest in 1974, placed special emphasis on the linkages between women’s status and fertility and stressed the urgency to eliminate discriminatory barriers in the spheres of education and employment.  These recommendation were further strengthened at the International Conference on Population held in Mexico city in 1984, The promotion of women’s education and well-being was also a priority theme at the 1994 International Conference on Population and Development held in Cairo. The relationship between education and fertility has been a constant theme in the demographic literature.  The availability of comparable data for a wide range of societies in the last three decades have also provided with a unique opportunity to elucidate the causal linkage between education and fertility (Cochrane, 1979; Jain, 1981).   It became evident that the relationship was largely contingent on the level of development, social structure and cultural milieu (Dyson and Moore, 1983).   Awareness of existing complexity of the relationship also stimulated in depth exploration of the underlying mechanisms through which education shapes the biological, psychological and social context of child bearing.

            Education has been found to delay entry into marriage, to  favour a normative orientation towards smaller families and to increase awareness, access and acceptability of contraception.    However, in the poorest and least literate societies small improvements in female education initially increase fertility, by improving maternal health and reducing the duration of breast-feeding and post partum sexual abstinence.  But once the process of child bearing becomes at least partially, subject to conscious planning, the relationship between female education and fertility is bound to be unequivocally inverse.

OBJECTIVE

            With the recent availability of the new round of National Family Health Survey (NFHS-II) data the goal of the present study tries to update the existing evidence on the role of women’s education in lowering fertility and increasing contraceptive use in various states in India.  Multivariate analysis (a number of OLS regressions) has been performed to disentangle the impact of women’s education from other socio economic characteristics.

DATA AND METHODOLOGY

Both NFHS-I and NFHS-II data are used.  However, only for eight states of India we had data of NFHS-II.   These states are Gujarat, Andhra Pradesh , Bihar , Madhya Pradesh, Orissa, Rajasthan, Uttar Pradesh and Haryana.   So, we have restricted the analysis to these states only. 

Several regression models are estimated for each state, based on individual data.  The effect of women’s education is assessed in relation to three outcomes:  Children ever born (representing actual fertility), desired family size (proxy for the demand factor) and conceptive use.  Two empirical models are presented for each dependent variable.   In model-I, women’s education is adjusted for demographic variables i.e. age at marriage, marital duration, marital duration squared.  In model 2, women’s education is adjusted for both demographic and socio economic variables (urban/rural residence, standard of living index, type of occupation and SC–ST/others).  The size of the regressions coefficients and their level of statistical significance are observed to measure the relative strength of education. The shape of the relationship can be ascertained by examining whether the coefficients increase in the linear and monotonic fashion. The comparison of regression coefficients in Model-I and Model-II permit the assessment as to what extent the observed impact of women’s education on fertility is attributable to socio-economic factors.

RESULTS AND DISCUSSION

Children Ever Born:

Table 1 presents the ordinary least squares (OLS) regression analysis of cumulative marital fertility i.e. children ever born. Although there are similarities across the states shown in Table 1, the magnitude and the pattern of the education-fertility relationship display certain distinctive features by states. In general, values of regression coefficients resulted from the analysis of both data (NFHS-I and NFHS-II) are linearly increasing and most of them are statistically significant (at 1 percent level). These results imply that women’s cumulative fertility can be substantially reduced by education up to an advanced level of schooling (High school+). The results i.e. the value of regression coefficients for different educational groups do not vary much between NFHS-I and NFHS-II for all the states. However, between the states there is a variation in the value of regression coefficients. For Gujarat the coefficient value is highest followed by Haryana, Utter Pradesh, Madhya Pradesh and Rajasthan. In the middle education category the coefficients are not significant for Andhra Pradesh, Bihar and for high school+, again for Andhra Pradesh the coefficient remained insignificant.

            The coefficients in Model-II (after adjustment for demographic and socio-economic variables) are drastically reduced for certain states like Haryana, Rajasthan and Madhya Pradesh (both data are indicative). However, for Andhra Pradesh and Bihar the effect of education is insignificant in the middle and upper educational strata. Individual education can be less effective in reducing fertility in context where overall education is low. In states where education is low the few women that have reached advanced schooling levels are likely to have high socio-economic status, which might affect the results (states like Rajasthan, Madhya Pradesh, Uttar Pradesh, Bihar). Conversely, the strongest effect of individual education is usually found in societies where women are, on an average, better educated.

Desired Number of Children:

            Table 2 presents the results from the OLS regression analysis on desired family size. In general, the effect of women’s education on demand for children is strong in all the states considered here. According to NFHS-II data, the effect of women’s education on the demand for children is found stronger (in comparison to the reference group which is illiterate) for Gujarat, Bihar, Orissa and Utter Pradesh (regression coefficients have higher value).

            After demographic and socio-economic factors are controlled, the effect of women’s education remains statistically significant for all states except for Uttar Pradesh in the case of primary and middle level educational groups. Although the effect of female education gets attenuated once controls are established, the effect remains strong In most of the states and displays a linear pattern. As per NFHS-II data, for Andhra Pradesh and Madhya Pradesh only an insignificant effect on fertility preference is observed, for the primary and middle level schooling groups in the case of former and for the middle level group in the case of  the latter. For Haryana, the effect is comparatively the weakest and non- linear.

Contraceptive Use:

Table 3 presents the results of the analysis of contraceptive use. Given the dichotomous nature of the dependent variable, a logit regression model is employed. The sample analysed is restricted to non-pregnant women currently in union. A comparison of the results of NFHS-I and NFHS-II data reveals that the association between female education and contraceptive use has intensified in recent years (the period between two surveys). As a result, the regression coefficients have turned to be significant for all states in the lower panel of Table-3 (refer to NFHS-II). The linearly increasing coefficients across the educational groups indicate the significant increases in the likelihood of using contraception. With no exception at all level of education, women having more years of schooling have considerably higher rates of contraceptive use than women with no formal education. The effect is higher in Madhya Pradesh and Rajasthan (almost equal after controlling for demographic and socio-economic factors).

SUMMARY AND CONCLUSION

The analysis revealed the negative influence of education on fertility. In all the states analysed here, fertility had declined with increased female education. Certain states (Bihar,A.P.) showed the influence of external factors(family planning programme) in lowering fertility. Reported differentials in desired fertility are much smaller than observed differentials in actual fertility. Study documented that better educated women consistently wanted smaller families. Education thus guarantees changing expectation and changing aspiration for children.

In many states unwanted fertility are high hence women are far from having achieved their reproductive goal. Unwanted fertility low among educated women suggests that education enables reproductive choice and reduced the gap between desired and actual fertility. Women’s education affects the ability and willingness to implement the fertility preference. We have seen strong relationship between education and contraceptive use.


REFERENCES

 

 

 

1.  Cochrane, Susan H. (1979). Fertility and Education. What do we really know ? Baltimore, Maryland: Johns Hopkins University Press

2.  Dyson, Ttim and Mick Moor, 1983. Kinship structure frmale autonomy and demographic behaviour inin India, Population and Development Review 9(1)

3.  Jain, Anrudh K. (1981). The effact of female education on fertility: a simple explation. Demography (Alexanadria, Virgina), Vol.18,No.4 (November), pp 577-595

4.  Kelly, Gail P., and Carolyn M. Elliott (1982). Women’s Education in the Third World. Albany, New York: State University of New York Press

5.  United Nations (1985): Report of the World Conference to Review and Appraise the Achievements of the United Nations Decade for Women: Equality, Development and Peace, Nairobi, Kenya, 15-26 July 1985. Sales No. E.85.IV.10.

6.  United Nations (1987): Fertility Behaviour in the Context of Development: Evidence from the World Fertility Survey. Population Studies, No.100 ST/ESA/SERA/100. Seles No. E.86.XII.5


 


Table 1:

OLS Regression Coefficients of the Effect of Wife’s Education on Children Ever Born

 

State

 

Wife’s Education

 

Primary

Middle

High-school +

 

 

NFHS - I

Gujarat

Model-I

-0.400*

-0.614*

-0.979*

 

Model-II

-0.363*

-0.540

-0.832*

Andhra Pradesh

Model-I

-0.257*

-0.200**

-0.391*

 

Model-II

-0.257*

-0.198**

-0.361*

Bihar

Model-I

-0.204*

-0.269**

-0.642*

 

Model-II

-0.138

-0.163

-0.384*

Madhya Pradesh

Model-I

-0.105

-0.282*

-0.662*

 

Model-II

-0.053

-0.166

-0.437*

Orrisa

Model-I

0.158*

0.112

-0.271*

 

Model-II

0.107

0.089

-0.162

Rajasthan

Model-I

-0.149

-0.275*

-0.637*

 

Model-II

-0.022

-0.079

-0.284**

Uttar Pradesh

Model-I

-0.358*

-0.438*

-0.925*

 

Model-II

-0.278*

-0.311*

-0.695*

Haryana

Model-I

-0.481*

-0.458*

-1.037*

 

Model-II

-0.332

-0.229

-0.687*

 

 

NFHS - II

Gujarat

Model-I

-0.367*

-0.763*

-0.960*

 

Model-II

-0.295*

-0.639*

-0.755*

Andhra Pradesh

Model-I

-0.178*

-0.131

-0.093

 

Model-II

-0.174*

-0.122

-0.040

Bihar

Model-I

-0.157*

-0.175

-0.336*

 

Model-II

-0.073

-0.066

-0.148

Madhya Pradesh

Model-I

-0.375*

-0.465*

-0.751*

 

Model-II

-0.274*

-0.289*

-0.464*

Orrisa

Model-I

-0.164*

-0.201**

-0.527*

 

Model-II

-0.240*

-0.266*

-0.514*

Rajasthan

Model-I

-0.364*

-0.584*

-0.713*

 

Model-II

-0.221*

-0.382*

-0.435*

Uttar Pradesh

Model-I

-0.352*

-0.514*

-0.831*

 

Model-II

-0.268*

-0.403*

-0.634*

Haryana

Model-I

-0.394*

-0.636*

-0.882*

 

Model-II

-0.206*

-0.341*

-0.401*

-Model-I controls for demographic variables (Age at marriage, Marital duration, Marital duration squared).

-Model-II controls for demographic variables and socio economic variables (Rural-Urban residence, Standard of living Index, Type of occupation and SC-ST/Others).

 

*p < 0.01;                 ** p < 0.05

 

Table 2:

OLS Regression Coefficients of the Effect of Wife’s Education on Desired Number of Children

 

State

 

Wife’s Education

 

Primary

Middle

High-school +

 

 

NFHS - I

Gujarat

Model-I

-0.318*

-0.598*

-0.704*

 

Model-II

-0.286*

-0.535*

-0.609*

Andhra Pradesh

Model-I

-0.267*

-0.389*

-0.416*

 

Model-II

-0.200*

-0.285*

-0.293*

Bihar

Model-I

-0.425*

-0.704*

-0.860*

 

Model-II

-0.272*

-0.512*

-0.577*

Madhya Pradesh

Model-I

-0.423*

-0.620*

-0.858*

 

Model-II

-0.286*

-0.411*

-0.565*

Orrisa

Model-I

-0.428*

-0.674*

-0.867*

 

Model-II

-0.239*

-0.413*

-0.557*

Rajasthan

Model-I

-0.459*

-0.523*

-0.759*

 

Model-II

-0.308*

-0.306*

-0.453*

Uttar Pradesh

Model-I

-0.445*

-0.603*

-0.827*

 

Model-II

-0.368

-0.475

-0.585*

Haryana

Model-I

-0.289*

-0.451*

-0.530*

 

Model-II

-0.204*

-0.329*

-0.379*

 

 

NFHS - II

Gujarat

Model-I

-3.451*

-4.110*

-5.158*

 

Model-II

-2.596*

-2.951*

-3.612*

Andhra Pradesh

Model-I

-3.678*

-6.514**

-7.744*

 

Model-II

-2.351

-4.824

-6.245*

Bihar

Model-I

-2.278*

-5.330*

-4.849*

 

Model-II

-2.202**

-4.810*

-4.087*

Madhya Pradesh

Model-I

-2.427*

-2.879*

-3.033*

 

Model-II

-1.824*

-1.952

-1.990**

Orrisa

Model-I

-2.326*

-2.147*

-2.664*

 

Model-II

-2.090*

-1.816*

-2.303*

Rajasthan

Model-I

-1.549*

-1.500**

-0.923

 

Model-II

-1.511*

-1.389*

-0.895*

Uttar Pradesh

Model-I

-4.631*

-6.747*

-8.704*

 

Model-II

-5.025*

-7.176*

-9.068*

Haryana

Model-I

-1.318

-2.111*

-0.812

 

Model-II

-0.737*

-1.242*

-0.658*

-Model-I controls for demographic variables (Age at marriage, Current age, Number of living children including current pregnancy).

-Model-II controls for demographic variables and socio economic variables (Rural-Urban residence, Standard of living Index, Type of occupation and SC-ST/Others).

 

*p < 0.01;                 ** p < 0.05

 

 

Table 3:

Logistic Regression Coefficients of the Effect of Wife’s Education on Current Use of Contraception

 

State

 

Wife’s Education

 

Primary

Middle

High-school +

 

 

NFHS - I

Gujarat

Model-I

0.1676

0.7974**

0.9535*

 

Model-II

-0.1310

0.4966

0.5333

Andhra Pradesh

Model-I

-0.6720

0.3023

0.6585

 

Model-II

-1.5493

-0.6168

-0.3121

Bihar

Model-I

0.1082*

0.2482*

0.5903

 

Model-II

0.2622*

0.3011*

0.6628**

Madhya Pradesh

Model-I

0.4027

0.5181

0.8615*

 

Model-II

0.1520

-0.1082

0.1930

Orrisa

Model-I

0.1469

-0.0961

0.8922

 

Model-II

-0.0501

-0.5318

0.2578

Rajasthan

Model-I

0.0963*

0.1876*

0.6924

 

Model-II

-0.2000

-0.2570

-0.2074

Uttar Pradesh

Model-I

0.4299*

0.1942

1.0800*

 

Model-II

0.2543

-0.1350

-0.5315*

Haryana

Model-I

0.4998

0.4097

0.7138*

 

Model-II

0.3631

0.1723

0.3247

 

 

NFHS – II

Gujarat

Model-I

0.4556*

0.6699*

0.8634*

 

Model-II

0.4559*

0.6176*

0.7642*

Andhra Pradesh

Model-I

0.5925*

0.7356*

0.6441**

 

Model-II

0.4449*

0.5156*

0.2991*

Bihar

Model-I

0.9182*

1.3032*

1.4305*

 

Model-II

0.5841*

0.8704*

0.7620*

Madhya Pradesh

Model-I

0.4948*

0.7844*

1.2878*

 

Model-II

0.3753*

0.5345*

0.8475*

Orrisa

Model-I

0.3072*

0.9510*

1.1583*

 

Model-II

0.1200*

0.6491*

0.6551*

Rajasthan

Model-I

0.5690*

0.8690*

1.3433*

 

Model-II

0.3085*

0.4886*

0.8340*

Uttar Pradesh

Model-I

0.5229*

0.8419*

1.3196*

 

Model-II

0.3602*

0.5908*

0.7960*

Haryana

Model-I

0.3280*

0.4793*

0.7896*

 

Model-II

0.1685*

0.2518*

0.4215*

-Model-I controls for demographic variables (Age cohort, Number of living children).

-Model-II controls for demographic variables and socio economic variables (Rural-Urban residence, Standard of living Index, Type of occupation and SC-ST/Others).

 

*p < 0.01                ** p < 0.05