August 2008 - Volume 2, Issue 4

The Effect of Women's Education and Some Socio- Economic Variables on Fertility and Contraceptive Use in Bangladesh: A District Level Analysis

Md. Abdul Gonia, b a IER, Hitotsubashi University, Kunitachi, Tokyo 186-8601, Japan and b Assistant Professor, Department of Population Science and HRD University of Rajshahi, Rajshahi - 6205. Bangladesh Email: magoni_popsrubd@yahoo.com

ABSTRACT The purpose of this study is to present the fertility and contraceptives levels in different regions of Bangladesh and to determine the most important determinants of fertility and contraceptive use of Bangladeshi women. It utilizes mainly the BMMS 2001 and Bangladesh District-Level Socio-Demographic and Health care Utilization Indications that cover all geographic areas of the Country. Using some statistical techniques such as cross tabulation, correlation coefficients, important relationships between fertility and several demographic, socio-economic, and spatial variables, are explored. In an attempt to understand fertility behavior of Bangladeshi women, a multiple regression analysis was utilized. It is found that woman's education and working status are apparently the most important determinants of fertility behavior and contraceptive use. It is also found that some variables, specifically child mortality, land ownership, household with TV, NGO involvement and the geographic region, are significant determinants of fertility and contraceptive use in Bangladeshi women. Finally, the study was able to present a few suggestions and recommendations. Key Words: Fertility, Contraceptive Use, Education, NGOs and District Level Data.

INTRODUCTION

Bangladesh is a country which like many other developing countries, has achieved a remarkable reduction in fertility despite little improvement in levels of living, education, women's status, child survival and other factors frequently associated with the demographic transition. The family planning program is credited with being the main driving force behind this reduction, while the role of social and economic change is de-emphasized (Larson and Mitra 1992; Cleland et. al. 1994). Bangladesh is internationally considered a success story in family planning (Freedman, 1995), with an increase in contraceptive prevalence rising from 8 percent to 54 percent and a decline in the total fertility rate from 6.3 to 3.3 in the three decades since independence (BDHS 2000). Success in meeting these population goals can be largely attributed to the commitment of the Government of Bangladesh (GOB) and the Ministry of Health and Family Welfare (MOHFW), which have effectively coordinated donor organizations, non-government organizations (NGOs) to ensure that free or affordable contraceptives are available in both public and private health facilities throughout the country. The late 1980s has seen a large increase in the number of couples using family planning methods. Unfortunately, the use of family planning has declined to 50 percent in 2001 BMMS. The decline in overall use is due to a decline in the use of traditional methods (from 10 to 6 percent). Use of modern methods has little changed since 1999-2000.

As for social indicators women have gained in the development process, levels of female labour force participation rose from the low level of 15.5% in 1995 to 23.9% in 1999-2000. The NGOs and the garment industry have absorbed a good number of the female labour force. Electricity is another important element of human life for developing countries like Bangladesh. Electricity generates employment. The impact on employment is both direct and indirect. Women in the electrified households are involved more in household level income-generation activities and depict better re-allocation of time for remunerative employment; unemployment rate is relatively low in the electrified households; and there is a relatively higher share of non-agricultural employment in the electrified households indicating a modernization effect of electricity on occupation. The overall literacy rates for both males and females in the electrified households are higher, especially due to the household's access to electricity which has contributed much both in economic terms as well as in raising awareness about value of education (Barkat, et. al. 2002).

Demographic and social-economic variations exist in Bangladesh across different regions. The total fertility rate is lowest in Khulna division (2.5) and highest in Sylhet division (4.3), which is 2 times higher than Khulna division. Women with no formal education have more children (3.8) than formal educated women (2.5) with at least secondary education. Like education, women in poorer households have more children than women in wealthier households. With a TFR of 4.2, women in the poorest households are likely to have about two children more than women in the wealthiest households (TFR of 2.4). Proportion of women working for cash/kind is lowest in Sylhet division (22.3) and highest in Khulna division (31.8) (BMMS, 2001). There have also been significant changes in the fertility and women employment related variables, like infant mortality rate, CPR, female education etc.

Apart from being of intrinsic interest, these interregional and inter-temporal variations provide useful opportunities to study the determinants of demographic outcomes of different districts in Bangladesh. This paper examines some of the relevant relationships between TFR and contraceptive use based on a cross-sectional analysis of district-level data from the Bangladesh Maternal Health Services and Maternal Mortality Survey (BMMS) 2001. This paper also tried to determine the most important factors affecting reproduction behavior. Specifically, objectives of this paper are: a) to present the levels of fertility and contraceptive use in Bangladesh; and b) to determine the most important variables that explain the fertility behavior and contraceptive use.

From a fertility point of view the fertility determinants in a population is a complex process. While fertility behaviour influences population growth, which has consequences for pressure on resources, employment situations, health and other social facilities, and savings and investment, in turn, such consequences have a great bearing on the socio-economic variables that affect fertility behaviour. Socio-economic conditions of a population and differences in them affect the level of fertility in a population and create differences among the sub-groups or sub-regions. The demographic and the socio-economic variables considered for investigating the fertility and contraceptive use by various sub-regions of a population are: education of women, work status of women, child mortality and household condition of women.

In this study, samples consist of 64 districts for which detailed information is available. Under the study the demographic outcomes are total fertility rate, contraceptive use and some socio-economic variables. Particular attention is paid to female literacy, female household conditions (electrified or not) and involvement in NGO activities.

This paper is organized as follows. In section 2 we discuss data and methodology, section 3 analyses the data regarding women's education and fertility; and presents the regression results, conclusion and finally offers policy implications in section 4.

DATA AND METHOD

The 2001 Bangladesh Maternal Health Services and Maternal Mortality Survey (BMMS) created an opportunity to address this information need. This study utilized mainly the BMMS 2001 and Bangladesh District-Level Socio-Demographic and Health care Utilization Indications. The BMMS is the first national survey conducted in Bangladesh to serve as a source of maternal health and maternal death data for policy makers and the research community. The survey also collected other information such as household socio-economic conditions, education, reproductive and child health. In the past, most data on the country's socio-economic development, health and family planning were only available at the division level. Due to the large sample size of BMMS, it is possible to present this information at the district level. The survey was implemented under the authority of National Institute of Population Research and Training (NIPORT) in collaboration with ORC Macro, USA. The Johns Hopkins University, USA and ICDDR, B provided technical assistance. Associates for Community and Population Research (ACPR), Mitra and Associates, two Bangladeshi private research firms, collected the survey data of all sample points from November 2000 to April 2001. Some other sources such as Bangladesh Demographic and Health Survey (BDHS) 2004 has been used.

Administratively, Bangladesh is divided into six divisions. Each division is divided into districts (Zilas), for a total of 64 districts and then thanas/upazilas. Each urban area in a Thana is divided into wards and each rural area in a Thana is divided into union parishads. In each division, the list of wards constituted the initial sample frame for urban areas and the list of union was the sample frame for rural areas. At the first stage, a total of 1,616 clusters from 808 (674 urban and 134 rural) primary sampling units were randomly selected. A second stage, a systematic sample of 104,323 households was selected to interview more than 100,000 ever-married women age 13-49.

Three types of questionnaires were provided for the BMMS: a Household Questionnaire, a Women's Questionnaire (for ever married women age 13-49), and Verbal Autopsy Questionnaire (for deaths of women age 13-49). The survey collected basic information such as age, sex, marital status and education. Information was also collected on household characteristic such as type of housing, sources of water and availability of electricity. The information about female adult deaths identified deaths for which the Verbal Autopsy Questionnaire was used.

The ever-married women age 13-49 were asked questions on the following topics:
Background Characteristics (age, education, religion, etc), reproductive history, use of family planning methods, information about siblings (to calculate the maternal mortality rate).

Fertility is measured by the total fertility rate (TFR), and its component age-specific fertility rates (ASFRs). The TFR is defined as the number of children a woman would have by the end of her childbearing years if she were to pass through those years bearing children at the currently observed age-specific rates. The child mortality is the probability that a child will die before attaining the age of five years.

Another indicator is level of education of women. Proportion of women aged 15-24 with completed secondary education was included in our analysis. Female labour force participation, included the percentage of women aged 15-49 participating in the labour force. I include two unconventional new variables, NGOs and electricity, which relate to women's employment and consumption of population.

Finally, region dummy variables are used to identify regional patterns. A list of the variables, their definition and sources are given in Table 1, which are the summary statistics of our data. Most of the information used in this analysis is derived from 2001 BMMS and another source is BDHS 2000, which are available in published survey reports (see Table 1).

In this paper two dependent variables are used: (a) Total fertility rate, and (b) Use of contraception to the BMMS 2001 data. Two categories are used for each of the two dependent variables: for the contraceptive use variable - use of modern family planning method and for the total fertility rate. The independent variables included were the woman's participation in the labour force, access to mass media, women's educational attainment (schooling, primary education and secondary education), and involvement in NGOs, landownership, electricity connection, and geographic region. The relationship between fertility, CPR and its determinants has been examined using some statistical methods such as Pearson correlation coefficient and multiple regressions for the district level data from BMMS 2001. In district level data, cross-sectional analysis is standard based on the assumption that the error terms are independently and identically distributed. If there is a possibility of correlation between error term and explanatory variables the ordinary least squares estimates will be inconsistent, and the estimated coefficients will not approach their true values even in very large samples (see Murthi, Dreze and Guio 1995). In order to test the consistency of our results two regression models were estimated; one with all independent variables (Model 1 and 3), and the other with only the significant ones (Model 2 and 4).

Sources: NGO: Bangladesh Demographic and Health Survey (BDHS 2000); and remaining information came from BMMS 2001

Several demographic, socioeconomic, and geographic variables are used for the analytical purpose (Table1). These variables represent some personal/household characteristics. Some of them are nominal in nature and included in the regression equations as dummy variables.

RESULTS AND DISCUSSION

The Bangladesh Demographic and Health Survey (2001-2003) provides some interesting information on fertility differentials (as measured by total fertility rate and the mean number of children ever born) and current contraceptive use and educational attainment of women. As expected, the educational attainment of women is strongly associated with fertility. Looking at the relationships between women's education and total fertility rate (Table 2); it appears that the TFR (2.2) of the women of at least secondary completed education is less than one half that of the no educated women (TFR 3.6). The TFR decreases with rising levels of education. Table 2 also presents data on the same women's wanted (desired) fertility (TWFR). The Table shows that all educational groups of women, except those with no education, want to have around replacement-level fertility (2.1) or lower. Women who have some primary school education report that they want to have 2.1 children, whereas those with at least secondary wanted to have only 2.0 children. The wanted fertility for women with no education, still have a desire for an above-replacement level of fertility. For all educational groups, the total fertility rate exceeds the total wanted fertility rate. The difference is greatest for the group with no education - the TFR exceeds the TWFR by 1.3 children - and decreases with education to 0.2 children for the most educated group. If we can interpret this difference as unwanted fertility, we can conclude that enabling women to avoid unwanted fertility, e.g., through better family planning services, would lead to fertility rates around replacement level for all women except those with no education.

A quite similar picture emerges when one compares the contraceptive use by education levels of women; only those who have completed at least secondary education are more likely to report a higher use of both the modern and traditional methods than other women. It is mentioned here that TFR and mean ideal number of children are lower among the women who have completed at least secondary education. Generally contraceptive practice increases sharply with education (BDHS 2000). But here worth noting is that the successive two BDHS (1999-2000 and 2001-2003) suggested that contraceptive use has increased among women with little or no education (8%); while this figure is only 2% among the women who had at least secondary education. Those indicators in Table 2 suggest that educated women not only have different fertility goals but also have their aspirations based on reality. Further indications of these different links between fertility, contraceptive use and education and some socio-demographic variables will be seen on the next section.

Sources: Bangladesh Demographic and Health surveys 2004 about 10,000 ever-married women aged 10-49 years (p 51, 66, 113).
Notes: 1 see table 1; 2 Rates are based on births to women age 15-49 in the period 1-36 months preceding the survey.

3.1. FERTILITY LEVELS BY SPATIAL, DEMOGRAPHIC, AND SOCIO-ECONOMIC CHARACTERISTICS

Fertility behaviour is a complex phenomenon that results from the interplay of various social, psychological and cultural patterns related to employment, child mortality, contraceptive use, level of education and socio-economic development. It is not feasible to explain all the factors of fertility transition in Bangladesh together. Therefore, we need to classify the district into groups in such a way that they are highly heterogeneous between groups and homogeneous within groups to better understand the fertility behaviour. In this paper we considered 64 districts and six geographic broad regions/administrative divisions. It is very useful to present fertility levels by important woman's characteristics, familial attributes, and geographic regions of the country.

First of all, the data clearly shows that there is a strong relationship between fertility, contraceptive use and others demographic, and socio-economic variables (Appendix table A). The Chittagong and Sylhet division exhibit total fertility rate from 3 to 5, while the rates are lower, generally around 3, in the Rajshahi and Khulna divisions. The Dhaka division exhibits average fertility rates. The Barisal division is the third lowest fertility rate regions. The Rajshahi district under Rajshahi division shows lowest fertility rate (2.0). The Cox's bazaar (5.0) and Sunamganj (4.8) district shows relatively highest fertility rate at the same time. Generally, there are higher fertility rates there than in Chittagong division (3.74) and Sylhet division (4.27), and lower acceptance of the family planning method especially among the women of Cox's bazaar district (20.70) and Sunamganj district (18.90). The fact that the women of Rajshahi (29.43%) and Khulna (31.56%) divisions are more involved in NGO activities and more active in the labour force (30.20% and 31.3% respectively) (BDHS 2000 and BMMS 2001) than remaining regions, may also provide them with greater reasons to control the fertility rate.

It may be mentioned that broad region-level comparisons and classifications may not be able to capture fully the extent of diversities among various indicators characterizing several facets of development. Considerable regional diversity in terms of social, economic and demographic characteristics prevails in Bangladesh. This is true not only among the broad region but also among the districts of the same broad region. In general, these striking variations among the region in the livelihood of the common people stem from various factors such as the level of literacy, female education, nutritional standards, child mortality, morbidity, employment, income distribution, etc., and their corresponding interactions. These differentials of demographic and socio-economic variables among the regions interest researchers to observe the cross-relationships between fertility, women's employment and with socio-economic variables in district levels except spatial region. Table 3 produces the results of Pearson correlation coefficient of TFR, CPR and with some socio-economic variables.

"**" Correlation is significant at the 0.01 level, "*" Correlation is significant at the 0.05 level, (2-tailed)

The correlation between female labour force and fertility is -0.54 and statistically significant which implies that greater female labour force participation rate tends to show more decline in fertility rate. Greater acceptance of family planning methods means that the greater control of fertility rate. The correlation between these two variables is -0.86 (not shown in table) and is highly statistically significant. But the correlation between CPR and female employment is positive and statistically significant which implies that working women are more likely to use family planning methods than others.

Child death is also found to be strongly related to fertility and CPR among Bangladeshi women. The correlation between child mortality and fertility shows a significant positive relation and with CPR in a significant negative relationship. It means that if levels of child mortality, fertility rate would be higher and CPR would be decreased and vice versa. This finding is largely supported by most macro and micro fertility studies in different developing countries.

The close relationship between education, especially female education and demographic change has clearly emerged in recent empirical studies. Education is found to be strongly related to fertility and CPR. Illiterate women tend to have a larger number of children than those with higher levels of education. As shown in Table 2, an illiterate woman was found to have four children on average, while a woman who obtained at least secondary education had two children. Similarly, with the higher level of men's education is the smaller number of children. Based on the Pearson correlation coefficients, it seems that women's education is more strongly related to fertility and CPR than men's educational attainment. The importance of each of these two explanatory variables and their contribution in explaining fertility behavior and CPR will be examined in the next section.

The correlation between households (HH's) with electricity and fertility and HH's with electricity and CMR is negative and statistically significant except TFR. This implies that the electrified households are relatively lower in fertility and child mortality rate. Again the correlation between households (HH's) with electricity and female labour force and households (HH's) with electricity and education is positive and statistically significant except FLPR. So electricity has a contribution to reducing fertility, child mortality and increases literacy rate and created employment opportunities.

The fertility seems to be related to landownership. The fertility rate is higher among the women of poorest households than household with land. The correlation between poorest households and fertility is positive and statistically significant. The correlation between households with land and fertility is positive but not significant. The landownership is negatively related to CPR but not statistically significant.

Mass media is strongly related to fertility and CPR. The correlation between TV and TFR shows a significant negative relationship. The correlation between TV and CPR shows a significant positive relationship.

The correlation between NGOs and fertility is negative but does not show statistical significance. Maybe it has an interaction effect with other variables. But the correlation between NGOs and CPR shows a significant positive relationship. It implies that NGOs have an awareness ability among women regarding small families as norm, with health, education and created employment opportunities.

3.2. SOME DETERMINANTS OF FERTILITY BEHAVIOUR AND FAMILY PLANNING METHOD

Despite the importance of descriptive analysis in the previous section, it is very important to determine the variables that explain fertility behavior and contraceptive use in Bangladesh. Utilizing multivariate analysis in which all independent variables are taken into consideration, our hypotheses can be tested. The results of multiple regression analysis show that most of the independent variables have a significant effect on fertility and contraceptive use. Few regression coefficients of some variables such as male education, NGOs involvement, and household with land, are not statistically significant on fertility. In order to test the consistency of our results two regression models were estimated; one with all independent variables (Model 1), and the other with only the significant ones (Model 2) for TFR and Model 2 and Model 3 for CPR. Apart from indicating the signs of the coefficients and whether they are statistically significant, table 4 and 5 makes it possible to assess the quantitative effects of different demographic and socio-economic variables on fertility and CPR by combining the given information with the mean values presented in Table 1.

Note: Values of 't' are shown in parentheses. * Significance at less than 1 percent; ** significance at less than 5 percent and *** significance at less than 10 percent. For definition of variables see Table 1.

Note: Values of 't' are shown in parentheses. * Significance at less than 1 percent; ** significance at less than 5 percent and *** significance at less than 10 percent. For definition of variables see Table 1.

Source: Bangladesh District level socio-Demographic and Health care Utilization Indicators 2001(p 4-9).
Notes: 1 Percentage of women aged 15-49 participating in labour force; 2 Child mortality (deaths per 1,000 children surviving to the first birthday).

The multiple regression results presented in Table 4 and 5 shows that the female education is of the most important variable. As education level increases, fertility decreases and raises contraceptive use significantly. The results of any schooling and higher education are expected in sign but primary education is not expected in sign. Data from the World Fertility Survey and the Demographic and Health Survey confirm the positive effect of education on reproductive behaviour (Schultz 1994; World Bank 1994). Clearly, those with schooling beyond the primary level have a higher contraceptive use and lower fertility than those without. We expected a strong negative relation between fertility and male education and positively with contraceptive use. The fact is that male education has been found not to have a significant effect on fertility and contraceptive use. This is probably due to its interrelationship with both female and male education.

Female labour force is another most important variable that explains fertility behaviors and contraceptive use. It has a negative and highly statistical significant effect on fertility and positively on contraceptive use in the country. Contraceptive use appears to be a significant determinant of fertility and female labour force indicating they are appropriate instruments. The sharp increase in contraceptive prevalence has led to an appreciable decline in fertility (Khuda, B., et. al. 2000). Working women are more likely to use contraception and have fewer children as compared to non-working women (CPS 1991).

It is also found that child mortality significantly affects the fertility positively and contraceptive use negatively. The infant deaths shorten the period before next pregnancy and consequently lead to more births. This result is expected since most, if not all, fertility studies found similar results in different parts of the world. It is worthy to note that infant mortality rate (IMR) has sharply declined in Bangladesh during the last decade, which might suggest a further decline in fertility.

As expected, higher levels of poverty are associated with higher levels of fertility and lower use of contraception. It is found that the women of households with poorest land have higher fertility than those women of households with land. There is a positive statistically significant relationship between HH's with poorest land and fertility and negatively with contraceptive use. There appears to be little or no effect of landownership status on fertility and contraceptive use.

Another important economic condition of household level is electricity. About one-third (32%) of the households in Bangladesh have electricity (BDHS 2000). HH's with electricity have a significant negative effect on fertility and positively on contraceptive use (this result is consistent with Khuda, et. al. 2000 and Barakat, et. al. 2002). Electricity not only contributes to declining overall TFR, but also contributes to a reduction in TFR among the poor. Electrification has contributed to the positive development on women's socio-economic status. Electricity has left a profound impact on women's mobility, participation in IGAs, decision-making, freedom in using income and savings, better utilization of credit, knowledge about gender inequality issues, household work plan according to convenience, changes in attitude in terms of reducing healthcare disparities, increase in overall years of schooling for both boys and girls, preference to send girls to schools, awareness of legal issues (as for example, marriage for girls at 18 and boys at 21), and awareness about the negative impact of dowry. Although, women in the non-electrified households are working inside and outside home, they have less control over utilization of their earnings, decision-making; and their level of awareness of fundamental rights is low (Barkat, et. al. 2002).

The mass media like TV has a significant effect on fertility and contraceptive use. Women's access to mass media, especially TV is associated with higher probability of contraceptive use, and lower fertility and mean number of children. It is found to have a significantly negative effect on fertility and positively on contraceptive use.

Another important unconventional new variable is NGOs, which has a significant effect on contraceptive use. The regression coefficient corresponding to NGOs shows an expected sign and is statistically significant. The fertility rate is lower among the women associated with the activities of NGOs. But the result is not statistically significant. This is probably due to the fact that the relationships between fertility and NGOs reflect the joint influence of some others; time varying variables (see our earlier work, Goni, 2007). The NGO sector is currently playing an important role in informal education program (figures for any schooling, also see Table1) and created employment opportunities, especially among rural poor women (BDHS 2000). Non-governmental organizations (NGOs) in Bangladesh, especially the Bangladesh Rural Advancement Committee (BRAC), have been famous for their non-formal primary education programs that run low-cost schools for the poor. Because poor children help in household economic activities, school timing is set in such a way that the poor can participate in both school and household economic activities. Non-formal schools emphasize girls' education by enrolling more girls than boys. Involvement of women with NGOs required traveling among different places to attend meetings and training, and deposit savings and credit installments to bank. All these factors exposed them to new ideas, knowledge and experiences through their interactions with the outside world, consequently reducing fertility and increasing contraceptive use among the Bangladeshi women (Goni, 2007).

In developing countries, especially Muslim majority countries some researchers have found that there is no effect of work status on fertility (Noor's, 1986). A weak relationship between work status and fertility in some developing countries is due to availability of child care through the help of relatives (Chaudhury, 1978; Zurayk, 1987). In addition, it is not necessary for participation in the labor force to always lead to reduction of number of children, because of the competition between bearing children and work, for the time of mother and father (Easterlin, 1975). But in my study, there is a relatively strong relationship between education level and woman's participation in the labor force. For this reason, an interaction term between education and female labour force was introduced into the regression model. It is found that this term is statistically significant, indicating that the effect of education varies based on work status. As for example, the effect of at least secondary education is a little less for working woman compared to non-working woman (-0.08 + 0.03 = -0.05). This result emphasizes the weak impact of "labour force" upon fertility behavior.

The geographic variables are significant in explaining the variations of fertility and contraceptive use in the country. Some dummy variables have been introduced to represent major geographic regions in Bangladesh. It is found that a woman living in Rajshahi and Khulna division has lower fertility rate and higher contraceptive use than those in the Central Dhaka division. In general, fertility rate is higher and contraceptive prevalence rate (CPR) is lower among the women of Chittagong and Sylhet division. Our findings show that the fertility rate is in decline and contraceptive use increases over the entire region except Sylhet division. But the results are not significant for Rajshahi and Khulna division. It is worth noticing that the Sylhet division fertility rate is increased and contraceptive use decreases. It is observed from Table 4 and 5 that the fertility rate is in decline significantly in high fertility regions, that are Chittagong and Barisal divisions, but the low fertility rate regions like Khulna and Rajshahi division fertility rate of decline is not pleasing and gas likely stalled in recent decades. This result is also consistent with the findings of other studies (SVRS 2001; Islam 2003). It also seems that geographic variables capture the effects of some variables that are not in the model.

The overall explanatory power of the regression model is satisfactory (R2 = 0.81 and 0.80). This indicates that the model was able to explain 81% of the variations in the dependent variable TFR and 80% CPR. In sum, it was shown that female education especially secondary completed, is one of the most important variables in the model. This means that fertility and contraceptive use in Bangladesh is influenced by female education, rather than the impact of landownership. In addition, some other variables are found to affect fertility and contraceptive use such as child mortality, female participation in the labour force, poverty, electricity and mass media.

CONCLUSION AND POLICY IMPLICATIONS

In many other developing countries, Bangladesh is one of the best examples of a country with a strong family planning programmed effort, which has brought a significant fertility decline. In an attempt to understand the levels of fertility in Bangladesh in general, and to determine the major factors affecting fertility behavior and contraceptive use, in particular, the data of a BMMS 2001 and Bangladesh District-Level Socio-Demographic and Health care Utilization Indications data were utilized and some statistical methods were used. The results of the analysis indicate the important role played by women's education in fertility decline and increased contraceptive use in Bangladesh. The main findings can be summarized as follows:

Women's education has long been recognized as another crucial factor that influences childbearing patterns. Women's education reduces her desired family size and increases contraceptive use. Total fertility rate (TFR) and total wanted fertility rate (TWFR) decrease with rising educational levels.

Regression analysis revealed that female educational attainment and female labour force participation are the most important variables in explaining fertility behavior and contraceptive use in Bangladesh. Other variables were also found to be significant determinants of fertility and contraceptive use such as child mortality, landownership, and household's asset like electricity, and TV. The NGO involvement and geographic variables are also significant in the regression equation. We found that the fertility rate declines significantly in high fertility regions such as Chittagong and Barisal divisions but the low fertility rate regions like Khulna and Rajshahi divisions, fertility rate declines and is not pleasing as it has stalled in recent decades.

Moreover, the findings of this study have important policy implications, especially in formulating national population policy and useful when addressing female's participation in the labor force. Our finding indicates that improvements in both education and family planning services should receive priorities in policies. Education is important for reducing fertility and (and also infant and child mortality), as well as in its own right for improving the human capital (and economic potential) of the population. There is need to give at least secondary education for all women to further accelerate the lowering of the fertility rate in the country. Family planning services can help women avoid unintended pregnancies and the abortions that sometimes follow them (Rahman and others, 2001). We find that there is a substantial amount of fertility that is excess of desired fertility. Excess fertility is higher among women with no or little education. Family planning programs can play a crucial role, especially among the women with no or little education, in reducing the gap between desired and actual fertility. Women's involvement in NGOs and participation in labour force are crucial factors for reducing fertility and using contraception. Therefore, the policy maker should carefully design strategies that with better counseling and supervision should lead to increases in contraceptive adoption and continuation and hence should further reduce fertility in the country.

ACKNOWLEDGEMENT

I would acknowledge the contribution of my supervisors Prof. Osamu Saito and Prof. S. Taniguchi for their help and advice. I benefited from discussion at Professor Saito's seminar in the Institute of Economic Research (IER), Hitotsubashi University. I am grateful to Professor Osamu Saito for very useful suggestions and language editing every stage of this paper that improved essentially the content of the paper.

REFERENCES

1. Bangladesh Bureau of Statistics (BBS). Report of the Sample Vital Registration System, 20002, Dhaka, Bangladesh, 2003. BANBEIS. National Education Survey (Post-Primary)-, Dhaka, 1999.
   
2. Barkat, A., Khan. S. H., Rahman, M., Zaman, S. Economic and Social Impact Evaluation study of the Rural Electrification Program in Bangladesh. Human Development Research Centre, Dhaka, NRECA International Ltd, 2002.
   
3. Chaudhary, Sohanvir S. "Television in Distance Education: The Indian Scene." In Indian Journal of Open Learning, 1992, 1(1): 23-31.
   
4. Chaudhury, R., H. Female Status and Fertility in a Metropolitan Urban Area of Bangladesh, Population Studies, 1978, Vol. 32, No. 2, 261-273.
   
5. Cleland, J., J. F. Philips, S. Amin and G. M. Kamal. The Determinants of reproductive Change in Bangladesh. Washington DC: World Bank, 1994.
   
6. Clealand, J. "The Determinant of Reproductive Change in Bangladesh, Success in challenging Environment Washington, D.C. World Bank, 1994.
   
7. Coombs, P. H. and Ahmed, M. Attacking Rural Poverty. How non-formal education can help, Baltimore: John Hopkins University Press, 1974.
   
8. Easterlin, R. A. "An economic framework for fertility analysis." Studies in Family Planning, 1975, 6(3):54-63.
   
9. Easterlin, R.A. and E.M. Crimmins. The Fertility Revolution, Chicago: University of Chicago Press, 1985.
   
10. Erica, F., and Ambrus, A. Early Marriage and Female Schooling in Bangladesh; working Paper, Harvard University, 2006.
    
11. Freedman, R.. Asia's recent fertility decline and prospects for future demographic change. Asia-Pacific Population Research Reports No. 1. Hawaii: East-West Center, 1995.
    
12. Goni, M. A.,. Fertility and Women's Employment: An Analysis of District-Level Time-Series from Bangladesh. Paper Presented an International Seminar at the Islamic Countries Conference on Statistical Sciences, 2007, 12-14 December, Kuala Lumpur, Malaysia.
    
13. Islam, N. Urbanization, Migration and Development in Bangladesh: Recent Trends and Emerging Issues, Demographic Dynamics in Bangladesh, 2003, 125-146.
    
14. Jeffs, T. and Smith, M. K. Informal Education. Conversation, democracy and learning, Ticknall: Education Now, 1996.
    
15. Khuda, B., Roy, N. C. and Rahman, D. M. M. "Family Planning and Fertility in Bangladesh", Asia- Pacific Population Journal, 2000, Vol. 15, No. 1, 41-54.
    
16. Khuda, B. Fertility decline amidst unfavorable socio-economic conditions: Bangladesh a test case, Paper presented at Workshop on the Development of long-term strategic plan for FPAB, Cox's Bazar, 1991.
    
17. Khuda, B. and M. B. Hossain. Fertility decline in Bangladesh: towards an understanding of major causes, Health Transition Review, 1996, Supplement 6, 155-167
    
18. Larson, Ann and Mitra, S. N. Family Planning in Bangladesh: An Unlikely Success Story, International Family Planning Perspectives, 1992, 18(4), 123-144.
    
19. Lucas, Robert E. "The Industrial Revolution Past and Future." The Region. Federal Reserve Bank of Minneapolis. Online. Available, 2003: http://www.minneapolisfed.org/pubs/region/04-05/essay.cfm. 2003.
    
20. Martin, T. C. and Juarz, F. The Impact of Women's Education on Fertility in Latin American: Searching for Explanations, International Family Planning Perspective, 1995, Vol. 21. No. 2
    
21. Mitra and Associates. Bangladesh District Level Socio-Demographic and Health Care Utilization Indicators. Dhaka National Institute of Population Research and Training (NIPORT), Dhaka, Bangladesh, 2001.
    
22. Mitra, S. N., Charles, L., and Islam, S. (1993). Bangladesh Contraceptive prevalence Survey-1991: Final report. Dhaka: Mitra and Associates.
    
23. Mitra and Associates. Banglasdesh Demographic and Health Survey 1999-2000. Dhaka, Bangladesh and ORC Macro Calverton, Maryland USA, 2001.
    
24. Bangladesh Maternal Health Services and Maternal Mortality Survey 2001. NIPORT, Dhaka, Bangladesh, ORC Macro, Calverton and Johns Hopkins University, Baltimore, Maryland, USA, 2003.
    
25. Bangladesh Demographic and Health Survey (2001-2003). Dhaka, Bangladesh and ORC Macro Calverton, Maryland USA, ,2004.
    
26. Murthi, M., A. C. Guio and Dre`ze, J. "Mortality, fertility and gender biase in India: A district level analysis," Population and Development Review, 1995, 21: 745-782.
    
27. Noor, A. H. "Fertility in Jordan and its social, Economic, and Physiological Determinants," Journal of College of Arts (King Saudi University), 1986, 13(1): 191-226.
    
28. Rahman, M., Julie, D. and Razzaque, A. Do better family planning services educe abortion in Bangladesh. The Lancet (London), 2001, vol. 358, no. 9287, pp. 1051-56.
    
29. Schultz, T. P. 'Demand for Children in Low Income Countries', in Rosenzweig and Stark, 1997.
    
30. World Bank. Partnership in Development: The World Bank Assistance to Bangladesh. Dhaka Office: World Bank, 1997.
    
31. Zurayk, Huda. "An Overview of the Development of Studies on Fertility," Population Bulletin of ESCWA, 1987, 31: 65-76.