Education and Women’s Empowerment in Bangladesh Nashid Kamal *and K.M Zunaid**

*Professor and Head, Department of Population-Environment Independent University, Bangladesh **Graduate student, Sydney

1

Abstract: In Bangladesh the latest BDHS, 2004 finds that only twenty percent women work for cash. Among them only 48 percent are able to spend their money on their own, the rest are dependant upon spouses or other family members to take joint decisions on spending. Defining this variable as a measure of empowerment, this analysis investigates correlates of empowerment using logistic regression analysis. The model finds that marital status is the most significant predictor of empowerment in Bangladesh. Unmarried women are six times more likely to be empowered. Secondary education is another major determinant of empowerment, along with the woman’s mobility and decision making scores. The paper recommends strengthening women’s secondary education in Bangladesh and also taking measures to encourage late marriage for women.

2

INTRODUCTION The participation of women in the labour market in Bangladesh has increased in the recent years. One of the major areas where women have found employment has been the garment sector. This industry employs around 10 million persons directly or indirectly, of which ninety percent are women workers. In the 1993-94 BDHS the percentage of women who reported cash income was only 14.4 percent. After the garment industries were set up in Bangladesh, in 1996-97 a major increase is documented in the percentage of women employed for cash (Figure 1). However, this percentage fails to increase, in fact it decreases to 19.2 percent in 1999-2000 DHS and finally the most recent BDHS (NIPORT, 2005) reports only 17.5 percent women working for cash. The rise and fall of women cash earners in Bangladesh has been simultaneous with the changes in the garment industries. After 12 percent growth in the year 2000, economists noticed a significant decline in garment exports in the year 2001. More than half of the factories had no orders. According to Ward et al. (2004) `Declines in garment orders accelerated rapidly after September 11,2001. By November 2001, two thirds of the factories had no work orders .The changes in 2004 due to Multi-Fiber Agreement (Green, 1998) resulted in throwing 800,000 to 1,000,000 of the workers out of jobs. This is reflected in the national survey where the current number of women workers form only 17.5 percent, resulting in only 21 percent increase since 1993-94. Table 1 presents the data from successive surveys and finds that women with no education and those with tertiary education are the ones who have highest percentages of employment for cash. The case of the uneducated women is self explanatory as they are representatives of the lowest wealth quintile and are forced to work for a living, accepting whatever form of work is available. Similarly, the high rates of employment among women with tertiary education is self explanatory. Cash employment statistics for women with primary and secondary education has either increased insignificantly or has decreased (BDHS,2004secondary level). Kamal and Haider (2006) have found that in the recent years more and more women with primary and secondary education have accepted sex work as employment and in fact their study show that out of other categories, the sex workers have the highest levels of literacy ! The HDI has soared from 147 in 1993 to 138 in 2004. One of the major sources of contribution has been the increase in the women’s participation in the labour force. The garment sector has had a lion share in contributing to the same. However, in other sectors too, the government has tried to increase women’s participation and has introduced some `women’s only’ jobs to encourage women’s participation. For example, in the ministry of agriculture, some positions for poultry farmers and vegetable croppers have been reserved for women only. Much earlier in the National Health and Family Welfare Ministry, female family planning workers were recruited to serve as `change agents’ (Kamal,1994). Both in the government and private sector, primary and secondary school teachers have been recruited from amongst women. Laudable NGOs have exemplified other kinds of work which were earlier inconceivable as women’s work. Road building, guarding tree plantations, microcredit loan for small time enterprise etc have created

3

further avenues of employment for women, specially with marginal resources as well as educational qualifications.

In Bangladesh, educational levels have also increased among women. From a mere 25.8 percent in 1991, the current literacy rate is 43.4 (BBS,2002). It is expected that as educational levels are enhanced, women will have increased agency as well as negotiating powers both at home and at the work place. Other covariates of empowerment for example socio- economic status, regional variations, religious affiliation may also have an effect on the empowerment status of women. In Bangladesh, women’s empowerment variable has been included in the DHS surveys since 1993 when the number of female workers had just started increasing. Successive surveys have documented this variable regarding women’s empowerment, recording her ability to spend her own income on her own or by taking joint decisions with husband or other family members. This variable has been considered as major indicator of a woman’s empowerment. However, there has not been any study which looks at the influence of women’s educational levels on her empowerment status.This article examines the effect of education on women’s empowerment in Bangladesh net of other socio-economic covariates which could determine women’s empowerment in Bangladesh. METHODS AND MATERIALS The study uses data from the 2004 Bangladesh Demographic and Health Survey (BDHS). This is a two stage nationally representative survey. The 2004 BDHS sample is a stratified, multistaged cluster sample consisting of 361 primary sampling units (PSU), 122 in the urban area and 239 in the rural area. The PSUs were borrowed from a sampling frame created for the 2001 census of Bangladesh and which was termed enumeration areas (EA).All eligible woman aged 10-49, from the selected households were eligible respondents for this survey. This resulted in a total of 11440 women completed the female questionnaire. Further details are available in Mitra et al (2005). Among the 11440 women were interviewed in the BDHS,2004, 2282 women reported working for cash or cash and kind. This variable was screened for missing data and a total of 2279 women were used in this present analysis. In this analysis, logistic regression results are presented using the software SPSS. The dependant variable here is the answer to the question, whether she is able to spend her income alone or makes joint decision with husband or any other member of the family. This variable is coded 1 , if she decides alone and 0 otherwise. A total of 2279 working women were selected from the BDHS, 2005 file. Of them, 48% decided alone on the spending and the rest had joint decisions. In the BDHS, there were many variables available for decision making and for the mobility of the woman. Following a technique used earlier by the author (Kamal and Sloggett,1993), a Principal Component Analysis (PCA) was conducted to reduce the number of variables and a decision making score and a mobility score was constructed for each woman. The method is explained in details in Appendix 1.The socioeconomic

4

index was already calculated in the BDHS, they were regrouped to form two categories one termed Low SES and the other High. Other variables such as religion of the respondent, her membership with any NGO, division to which she belonged, her educational levels, marital status, husband’s education and type of residence were entered as independent ones in the model. These independent variables were tested for significance using bivariate techniques such as chi -square tests. RESULTS The logistic regression was conducted with the dependant variable as the woman’s decision regarding spending. It is coded as 1 if she decides alone,0 otherwise (meaning joint with husband or some other member of the family).The distribution of the variables considered here are presented in Table 2. The final parsimonious logistic regression model is presented in Table 3. The regression was conducted by using stepwise regression, including and excluding variables at various stages according to their relative importance in the model (using loglikelihood values as the decisive indicator). The first variable that was entered was the woman’s own education. For this variable, `no education’ was the reference category and women with primary education differed significantly from the reference category, otherwise all other categories were insignificant. Women with primary education showed lower probability of being empowered, compared to women with no education. As a second variable, mobility score was entered into the model. This variable was highly significant as a predictor of empowerment in this model. As mobility increased, woman’s empowerment increased significantly at all levels. Decision making score was entered in the next stage. Those with low decision making score showed significantly lower levels of empowerment. Socioeconomic score (SES) was entered into the model. Those in high socioeconomic status had 70% higher probability of being empowered. Division of residence was entered into the model. Women from Rajshahi had significantly lower probability of being empowered. However, the SES variable lost its significance. Hence, the interaction term between SES score and Division of residence was entered into the model. It was found that those from Barisal and high SES showed significantly lower probability of being empowered (62% less), compared to those from Dhaka division and in SES score high. The term SES on its own loses its significance, indicating that the interaction term was a more significant predictor variable than SES alone. Type of residence was entered into the model and its effect was insignificant, same for the age of the woman. Both these variables were deleted from the model. Religion was entered into the model. Those from Islamic faith had lower probabilities of being empowered, compared to those from `Others’ religion. They had 62% lower probability of being empowered.

5

Marital status was now entered into the model. When this variable enters, it produces most significant changes in the model The odds ratio is highest compared to other variables. Women who are unmarried have almost 6 times higher probability of being empowered. It is interesting to note that the role of woman’s education changes. The primary group loses its significance and the secondary group is now significantly higher than the `uneducated’ reference category. Husband’s education was entered in the next step, interaction term of husband’s education with marital status showed no significance. DISCUSSIONS AND CONCLUSION This model finds that a woman’s marital status is the most significant predictor of her empowerment. Unmarried women are much more likely to spend the income on their own. The results match another study by the author where only around 300 women from four different categories of employment were asked about their control on spending. Rural –urban differences and divisional differences could not be studied in the small sample which was also involved women from the same wealth quintile. In this study when many more covariates have been made available and the sample size is nationally representative, exactly same findings are observed as the earlier small sample study (Kamal and Haider, 2006). In this BDHS data (NIPORT,2005) used for the analysis the mean age at marriage was reported to be 14.8 years ! Marriage is universal in Bangladesh. The age at marriage in Bangladesh remains unusually low, in spite of various efforts to educate women and ensure their workforce participation. In comparison with neighbours like India, where the mean age at marriage is currently 20 or with Muslim states such as Pakistan (age 22), the mean age at marriage is remarkably low (see paper on Age at marriage). In fact, it has declined slightly since the BDHS 1999-2000 (reported 15). However, on detailed examination, it is found that amongst those who have attained secondary education or higher the mean age at marriage is 19.8 years for a cohort currently aged 25-49 which is very encouraging in the sense that it indicates that education may work in delaying marriage (NIPORT,2005). Additionally the proportion of employed women who work for cash only is markedly higher for those who have completed secondary education, although the numbers were very small (ibid). The multivariate analysis used in this paper also finds that after controlling for other covariates, the effect of secondary education is extremely significant on her status of empowerment. This result supports us to strengthen pathways to women’s education and also supports the proposition that educated women are more likely to marry later and also more likely to have higher agency in their lives. In the current years, aided by various policies and incentives introduced by the Government of Bangladesh since 1993, Bangladesh has witnessed a rising rate in girls’ enrollment at school. In fact, school attendance is higher for girls compared to boys for ages 6-10 and 11-15.For those aged 16-20, school attendance drops sharply from levels seen at younger ages, and attendance is higher for males than for females (ibid). The

6

major reason for school drop out at these ages for girls has been `to get married’ and policy makers have argued how the society could work towards shifting the ages of marriage in order for the woman to complete her secondary schooling (Amin et. al, 2006).Creation of greater employment opportunities for women, social movement against the dowry system, strengthening the value of the girl child in the society and increasing incentives for completing secondary education for women have been identified as major pathways towards attaining this goal (ibid). In support of the girl’s secondary education, the GOB has already introduced policies for free education at the secondary level. These policies have been introduced in 2005 and there will be a time lag for the policies to produce more women with higher than secondary education, which in turn may produce a later age at marriage and more women in the employment sector. In the meantime, other correlated factors, for example publication of textbooks on time, provision for girl’s hostel in various remote locations, increasing teacher’s remuneration, problems with supply of electricity need to be dealt with. In support of a social movement against the dowry, the Prime Minister of Bangladesh has issued a memorandum in late 2003 urging all government officials to work towards abolishing the practice of dowry. However, reports of dowry deaths are still common and evidence is presented in everyday coverage of the news. Even though the parents realize the importance of continuing education, Amin et al (2006) explains that `the motives for marrying daughters off at an early age in order to save on dowry costs remain strong’. Amin et al. (2006) further state that `Early marriage among females typically results in large age differences between spouses because male age at marriage is considerably higher. In a setting such as Bangladesh, where age confers authority and status, it is reasonable to argue that large age differences contribute to women’s subservient status in conjugal life’ (2006). In this paper, truly enough, the detailed logistic regression analysis, controlling for other covariates of empowerment finds the `marital status’ of the woman to be the most significant predictor of her empowerment, unmarried women more at liberty to spend their own income. If we assume that this single variable `whether the woman is allowed to spend her cash income according to her sole decision’ completely captures the woman’s empowerment status, then the subservient status of married women is obvious from the results representing national level data. This is further strengthened in the model by a variable which defines `religious affiliation ‘and finds that women from Islamic faith have lower probabilities of being empowered. The `subservient role’ of women is much publicized in Islam and in this connection, the policy makers need to involve the religious leaders in their awareness campaigns. More BCC needs to be designed to support the working woman’s role and encourage her to be a valued partner. Bangladesh has already overcome vehement opposition from fundamentalists who have been rampant against some women’s employment schemes introduced by various NGOs in the early 90s () . However, with years of fruitful BCC and consistent campaigns these barriers have now been removed. It is therefore hoped,

7

that the women’s status within the family and her working women image can also be established in Bangladesh in a similar manner. In this connection, the media has a very important role in portraying the woman to an average viewer. More attention on promoting working women images through dramas, advertisements, interviews and feature films is recommended. Many women have been successful entrepreneurs and their successes and feats need to be shared nationwide in building up positive images not only for women from the elite group but also in portraying how women can contribute from grassroots level. Creation of female `icons’ by highlighting success of garment worker, female worker in agriculture, politics, health care etc are also some ways in which the traditional gender roles need to be addressed. As Basu (1996) writes `direct intervention from the mass media, for example, as is already being fostered by the rapid globalization of the economy and the culture in this region (even illiterate villagers now know about the benefits of Ariel washing powder, which even educated Indians had not heard a few years ago)’ can be appropriately put into the context of Bangladesh as well. The employment of female workers in the garment sector has already created an enhanced image for working women, further employment opportunities need to be introduced. As garment work is facing challenges of post multifibre agreement, further areas of female employment need to be evolved. The Government should consider additional employment facilities for women in areas such as drug industries , processed food and the like. Apart from training for traditional skills of sewing and cooking, the government can think of developing other skills for women. One organization titled Sathi has had positive results in training women for computers and car driving in Bangladesh Ward et al.(2004).These experiences may be replicated. The women workers in the health sector have also been a successful addition to the work force (Kamal,1994). Their contribution should be acknowledged and institutes should be set up for further training, which should prepare them for jobs in the local as well as international markets .Women from Phillipines , Indonesia, Srilanka and India have additional qualifications of knowing the English language and other languages apart from their skill training. The GOB has already planned to set up computer and language centres in various subdivisions. Women’s participation, enrollment and continuation needs to be supported in this direction. Additionally, introduction of creche facilities, maternal and paternal leave, flexi hours are some of the ways in which countries like Singapore and Malaysia have encouraged their women to face the challenges of the job market. Bangladesh is essentially a Muslim society and freedom of movement outside the homestead is of special interest because of its association with purdah which may be a deterrent from women seeking jobs outside their homestead. This paper finds that women who are more mobile ie they are able to shop alone, go outside village alone, visit health centre alone without being accompanied by a male relative; have higher odds of being empowered. The odds of being empowered increases with her degree of mobility. As Cleland et al. (1996) explains, these variables and what they capture may `also have a profound influence on mentality and outlook by severely circumscribing interactions with

8

the outside world and exposure to new ideas and models of behaviour’. Decision making also emerges as an important predictor of empowerment. The education of the woman, her autonomy and her decision making role all add up in the same direction. As Basu (1996) argues `..female schooling does seem to increase several aspects of female autonomy which (again, besides being good in themselves)’ are also useful ways to strengthen the role of women, in this case her levels of empowerment. The data finds that women from Barisal division, who are from high socioeconomic subgroup have lower probabilities of empowerment compared to women from Dhaka division and same SES status. This indicates that being from higher socioeconomic group does not indicate that the woman will be more empowered. There exists divisional differences, perhaps to be captured further by other variables such as level of female literacy in the division, number of women in the work force, employment opportunities for women in that division, the period of exposure to women’s full time employment, religiosity of the community.(variables at the community level). The urbanization in Bangladesh can be called the `primate’ type where growth has been centred around one city and division alone (Afsar,2000). By year 2020 Dhaka city is expected to be the second most populous city in the world, Tokyo being the first (). All the industrial and infrastructural growth has been centered around Dhaka city. This paper indicates that there is a strong need for decentralization, with special attention given to Barisal division, which is closest to the Bay of Bengal and most frequently suffers the consequences of environmental disasters. The major implication from this paper is therefore that female education is a necessary goal, and it needs to be pushed even more vigorously to cross the threshold of secondary education beyond which the net effects on female autonomy are all positive. Finally, as secondary education is being strengthened, Bangladesh would like to see more women speak out like Rupali, that they are able to sustain themselves from their own work. `Rupali has worked ever since her parents moved to Dhaka from their native village. As a child she collected wood and paper, she started garments work at 10, but lost her job when the foreign inspector came. Instead of school, she returned to flower sales in the streets, until her parents married her at a young age and she returned back to garment’s work. Her wages improved her situation at home with her husband, because `I did not have to rely on my husband because I had my own income. I could live with that. Husbands do not give any importance if we do not work’ (Ward et al. 2004).

9

Fig. 1 Distribution of women in cash employment in Bangladesh DHS,19932004.
30 25 20 15 10 5 0 1993-94 1996-97 1999-00 2004 14.4 25.5 19.2 17.5

10

Table 1: Distribution of women in employment by their educational levels, Bangladesh DHS 1993,1996-7,1999-00,2004. Education % in cash employment 16.4 None 1993 12.5 Primary 7.9 Secondary 25.4 Tertiary Total 1996-1997 14.4 26.1 20.5 15.6 32.4 23.2 23.9 16.6 13.8 27.7 19.8 25.9 17.3 12.9 22.3 20.0

None Primary Secondary Tertiary

Total 1999-2000

None Primary Secondary Tertiary

Total 2004

None Primary Secondary Tertiary

Total

11

Table 2: Distribution of women who earn cash for a living according to various covariates of empowerment, Bangladesh DHS 2004.
Variable Spends her own income Alone Jointly with others Her education None Primary Secondary Higher Mobility score None Low Moderate High Decision making score Low High SES* Low High Type of residence Rural Urban Geographical location Dhaka Chittagong Rajshahi Khulna Barisal Sylhet Marital status Unmarried Married Religion Islam Others Frequency 1095 1186 Percent 48 52

1157 595 376 153

50.7 26.1 16.5 6.

786 577 446 472

34.5 25.3 19.6 20.7

1158 1123

50.8 49.2

1373 908

60.0 40.0

1361 920

59.7 40.3

542 322 667 374 187 189

23.8 14.1 29.2 16.4 8.2 8.3

407 1874

17.8 82.2

1997 284

87.5 12.5

*From the original SES the first three values poorest, second and middle were merged to form the group`low SES’ and the other two groups fourth and richest were merged to form `high SES’.

12

Table 3 Logistic regression of Empowerment of woman on various covariates, BDHS 2004 Variables Odds ratio Significance Woman’s education 1.00 Nonea ns .831 Primary