JICA-RI WP No.154
JICA-RI WP No.154
JICA-RI WP No.154
No. 154
June 2017
Use and dissemination of this working paper is encouraged; however, the JICA Research
Institute requests due acknowledgement and a copy of any publication for which this
working paper has provided input. The views expressed in this paper are those of the
author(s) and do not necessarily represent the official positions of either the JICA
Research Institute or JICA.
By
Minhaj Mahmud *, Keijiro Otsuka †, Yasuyuki Sawada ‡, Mari Tanaka § and Tomomi Tanaka **
Abstract
We analyze the factors and dynamics that contributed to the empowerment of women in
Bangladesh. We first investigate the role of non-farm sector growth in facilitating female labor
force participation and educational attainment, and then we explore how women’s
decision-making roles in a household have improved over the same time period. Our results
indicate that the proportion of village non-farm labor force participation is positively
associated with female school enrollment as well as other indicators of women empowerment.
Moreover, microcredit participation is found to be associated with larger roles for females in
making household decisions particularly on non-farm activities.
*
Senior Research Fellow, Bangladesh Institute of Development Studies, and Visiting Scholar, JICA
Research Institute
†
Senior Professor, National Graduate Institute for Policy Studies (GRIPS), and Professor, Kobe
University
‡
Professor, Faculty of Economics, University of Tokyo, and Former Visiting Researcher, JICA
Research Institute. Corresponding author. Email: sawada@e.u-tokyo.ac.jp.
§
Assistant Professor, Hitotsubashi University
**
PhD candidate, GRIPS
This paper has been prepared as part of a JICA Research Institute project entitled “Empirical Study on
Risk and Poverty in Bangladesh.” All views and interpretations expressed in this paper are those of the
authors and not necessarily those of the supporting institutions.
1
1. Introduction
One of the UN Millennium Development Goals (MDGs) is to “promote gender equality and
empower women,” and the UN Sustainable Development Goals (SDGs) encourages the
continuous efforts to “achieve gender equality and empower all women and girls.” The
empowerment of women is important beyond its intrinsic value. Since women now represent 40
percent of the global labor force (World Bank 2012), enhancing women’s productivity by
improving their abilities and opportunities is necessary to make the labor market and resource
allocation more efficient. Women also play a significant role in effective resource allocation in
households, including for child’s health and educational outcomes as shown in previous studies
Recent studies highlight the progress that Bangladesh has made toward empowerment
in the last few decades (Alam 2012; Heath and Mobarak 2015; United Nations 2014; World
Bank 2012). 1 Empirically, the issue of women empowerment has been approached in a variety
of ways. One of the approaches is to measure the degree of women’s empowerment by using
relatively simple and measurable indicators such as women’s age at marriage, female secondary
allocation by gender, and women’s self-esteem and other psychosocial measures (Abadian 1996;
Alam 2012; Anderson and Eswaran 2009; Field and Ambrus 2008; Mahmud et al. 2011). For
example, school enrollment (the proportion of children aged 5–19 attending school) in
Bangladesh increased among girls from 33 percent in 1991 to 56 percent in 2005. In addition,
since its independence in 1972, the country has experienced a reduction in fertility from almost 7
to just over 2 children per family coupled with a considerable increase in labor force
1
The concept of women empowerment as well as indicators to measure it have been widely discussed
and studied by a number of researchers. As the concept and its indicators are highly influenced by
culture, history, and social norms within a country or community, there is no one-size-fits-all strategy to
examine every aspect of the issue.
2
participation among young women. In this paper, we follow the above approach and employ four
school enrollment; 2) female labor force participation; 3) marriage age (or probability of
marriage); and 4) fertility. Using a unique nationally representative household panel data set
covering the years 1988 to 2008, we analyze the factors and dynamics that contributed to the
Since the 1980s, Bangladesh has achieved rapid industrialization based on the
industries. This development has provided women, who used to be housewives with no earned
income, with opportunities to enter the labor force and earn a decent income (Amin et al. 1998;
Kabeer and Mahmud 2004; Rhee 1990). Although the development of non-farm sectors (Amin
and Sonobe 2014; Mottaleb and Sonobe 2011; Rhee 1990) and the nationwide indicators of
achievements in women empowerment (United Nation 2014; World Bank 2012) have been
widely investigated in the existing literature, there is a lack of empirical evidence on the
The increases in female non-farm job opportunities and the resulting improvement in earnings
could lead to increased investments in human capital for girls. For example, Heath and Mobarak
(2015) find that the development of the garment industry had a positive impact on female labor
force participation as well as female years of schooling, which in turn delayed childbearing and
female marriage in the urban areas in Bangladesh. However, they exclusively focus on the
development of the garment industry in restricted urban areas near the capital city, Dhaka.
Rather than being confined to a single sector such as the garment industry, our study uses
nationally representative village household data, examining the overall nexus between farming
and non-farm sectors, in order to investigate the role of growth in non-farm sectors as a whole on
3
Our results indicate that the proportion of women participating in the non-farm labor
force is positively associated with female school enrollment as well as some other indicators of
the empowerment of women. Looking at the broader set of non-farm occupations using
household data, we find that an additional year of education is associated with 0.18 unit increase
in the labor force participation in the non-farm sector and with 7.5 percent reduction in number
of childbirths.
We then investigate how women’s ability to make decisions within their households has
improved over the two decades and if so how this is associated with microfinance lending. Since
women were regarded as being more credit constrained but more reliable in terms of loan
repayments than men, rural women have been targeted by microfinance institutions (MFIs). The
rapid penetration of microcredit programs in rural areas of Bangladesh is often cited as a key to
income-earning, market-based activities. This could empower women since the lack of women’s
gender-based division of labor restricting women’s link to market-based activities (see Boserup
1970; Sen 1992; Kabeer 1997; and Duflo 2012). Moreover we assume that enhanced women’s
factors, which we investigate in the first half of this chapter, such as female schooling, female
labour force participation (FLFP), marriage and childbearing outcomes. Existing studies,
however, show mixed evidence on the impact of microfinance on women’s empowerment. The
only study with a clean identification strategy known to us is by Banerjee et al. (2015). 2 The
study found no prima facie evidence that microcredit access leads to important changes in
2
Early studies on the impact of microfinance on the empowerment of rural women such as Hashemi et
al. (1996), Kabeer (2001), Pitt and Khandker (1998), Pitt et al. (2006), Osmani (2007) and Schuler and
Rottach (2010) did not employ sufficiently rigorous empirical strategies.
4
intra-household decision-making capacity of women or in social outcomes. Yet, the scope of
their analysis is confined to an urban rather than rural population. As such, the overall role of
microfinance in empowering the majority of microfinance clients living in rural areas deserves
rigorous investigation. Using panel data we show that microcredit participation significantly
Bangladesh. We assume that the development of the non-farm sector enhances female
participation in earning activities and encourages female education, which results in the rise of
female school enrollment. The increase in female human capital is positively associated with
the probability of female labor force participation and the decision-making ability of women,
while negatively associated with fertility and the probability of female marriage. The decline in
fertility and the probability of female marriage could be caused by two factors. One is the
increase in the opportunity cost for women to marry early and have children early, and the
other is due to the improved decision-making ability of women on these issues. Although the
issue is quite important, the underlying cause of these changes is beyond the scope of our study.
As shown in Figure 1, we test a hypothesis that access to microfinance has a positive impact on
female decision-making ability and women’s empowerment. The other information in Figure 1
is provided for supplement purposes. We acknowledge that the development of MFIs has
education. It is also notable that the conditional cash transfer program, called the Female
Secondary School Stipend Program (FSSSP), launched by the Bangladeshi government in 1994,
5
has contributed to improvements in female secondary school education (Hahn et al. 2016;
We employ a panel data set, “Livelihood System of Rural Households Panel Data,” consisting
sampling method was adopted for the sample selection of 62 villages in 57 districts (out of 64
districts in Bangladesh) for the benchmark survey completed in 1988 (see Figure 2 for the
location of selected villages). The sample is nationally representative as shown by the previous
study for which data are available from official statistics (Hossain and Bayes 1994). The
repeated panel household survey was conducted with the assistance of the International Rice
Research Institute (IRRI) and the International Food Policy Research Institute (IFPRI). The
panel data covers information from 1,240 randomly selected households in 1988, 1,880
To grasp the overall picture of the data set, Table 1 summarizes the basic statistics of
selected variables. As we can see from the 1988 data, the average female population is quite
young having a low level of education. At the time, non-farm labor employment opportunities
were limited for women and labor force participation was low. Yet these indicators seem to
improve over time. In particular, fertility rates among very young women have declined
substantially from 0.63 in 1988 to 0.38 in 2008. Also, reflecting the rapid expansion of
3
The panel data were collected by late Dr. Mahabub Hossain. We gratefully acknowledge his
contribution and support for using the data set.
6
Schooling
Schooling outcomes can be measured by two variables: average years of schooling and school
enrollment rates for each age group, as shown in Figures 3 and 4, respectively. We observe
distinct patterns of schooling specific to gender in Figure 3: while men’s average years of
schooling has increased slowly, women’s years of basic education on average have increased
dramatically, suggesting that women caught up with men rapidly in terms of human capital stock.
This overall growth in the stock of education can be achieved by increasing investments in
education for each age between 6 and 18 years old (Figure 4). The average school enrollment
rate for boys’ increased from 1988 to 2000, but there has been little improvement in secondary
enrollment rates from 1988 to 2000 at both primary and secondary levels.
The second core variable is a binary variable of female labor force participation (FLFP) in the
non-farm sector as well as other employment statuses such as farmer; housewife; unemployed;
and student. Aggregating these binary variables, we can compute primary occupational
compositions which are shown in Table 2 for the younger age group (15–19 years old) and Table
3 for the middle age group (30–35 years old). To examine the role of village-wide exposure to
the non-farm sector in facilitating female labor force participation, we compute the percentage of
those who are part of the non-farm labor force out of total labor force in each village. Then, we
identify “high non-farm exposure villages” and “low non-farm exposure villages” which are
defined as the villages whose percentage of the non-farm labor force in a village is in the top 25
percent and bottom 25 percent, respectively, among our sample 62 villages. Tables 2 and 3
show the occupational composition of women and men in high and low exposure villages in each
age group (15–19 years old and 30–35 years old, respectively).
7
According to Table 2, the proportion of women aged 15–19 identified as housewives
decreased and those as students increased substantially in both the high and low non-farm
exposure villages. This transition from housewives to students among the young women seems
to be associated with delayed marriage and increased labor market participation. In the high
non-farm exposure villages, the proportion of non-farm occupations dropped slightly, but the
In contrast, among the older age group of 30–35-year-olds, there is neither a clear sign of a
declining proportion of housewives nor an increase in female labor force participation (Table 3),
which implies that the effect of non-farm sector development on female schooling and work is
particularly strong among young women and that the impact is weakened as women age.
Age of Marriage
The third variable of our concern is age of marriage. Early marriage has been a huge problem in
Bangladesh. In fact, the country has one of the highest rates of marriage among girls under 19
years old. Figure 5 shows the proportion of married males and females in each survey year by
age group. We note that almost all the Bangladeshi women who are older than 25 years old were
married. An important observation is that the early marriage at age of 15–19 and 20–24 remains
marriages, we compute the proportion of married women aged 15–19 and 20–24 in high and low
non-farm exposure villages (Table 4). As can be seen, high exposure to the non-farm sector
coincides with a lower percentage of early marriage among women ages 15–19 and 20–24.
Fertility
The fourth outcome variable of interest is the average number of children under the age of 12 per
woman between 15 and 35 years old, which is clearly declining (Table 5). Since our data set
8
lacks information on migrants and it is possible for children over the age of 12 to have left their
home village and migrated to urban area for study, we construct the fertility variable based on the
number of children under the age of 12, who are unlikely to leave their home town and live
separately from their parents. Figure 6 shows that this declining fertility is a common
phenomenon among different age groups. Furthermore, as shown in Table 6, higher exposure to
the non-farm sector seems to be correlated with a more rapid decline in fertility.
Figure 7 shows changes in the decision-making patterns in a household, indicating that the
proportion of decisions in livestock production and fruit production made by women increased
between 2004 and 2008. Figure 8 shows the decisions made by men, women, and jointly, using
households. We find that the proportion of joint decisions over livestock and fruit production is
larger among microfinance clients. As to the vegetable production, the proportion of female
Empirical Strategy
In order to investigate a nexus among non-farm sector development and women’s empowerment
captured by investments in female education, female labor force participation, marital status,
and fertility decisions, we adopt the canonical regression methods using household panel data.
More specifically, we employ four different regression models: First, a school enrollment model;
second, a female labor force participation model; third, a marriage model; and finally, a fertility
decision model. Figure 9 summarizes the interrelationship among these models with relevant
variables. The left equation in Figure 9 examines the determinants of female school enrollment.
The non-farm sector development is our focal variable, and we control other factors such as
9
parents’ education, land ownership, the number of siblings, development of infrastructure, and
distance from major cities. Then, our analysis moves to the right-hand equations that examine
the impact of female years of education on other women empowerment indicators such as the
FLFP dummy, marriage dummy, and fertility. Combining results gained from the two sets of
equations, we can gain a deeper understanding of the impact of the non-farm sector growth on
Regression Models
We assume that years of schooling, H, and annual school attendance, S, are related as follows:
where 𝐻𝐻𝑖𝑖𝑖𝑖 denotes years of schooling of person i, 𝑎𝑎𝑖𝑖𝑖𝑖 is school attendance of person i in class T,
and 1[.] is an indicator function which takes one if the statement is true; and zero otherwise.
We then follow Estudillo et al. (2009) to specify the following linear probability model
(3) 𝑆𝑆𝑖𝑖ℎ𝑙𝑙𝑙𝑙 = 𝛼𝛼0 + 𝛼𝛼1 𝑎𝑎𝑎𝑎𝑎𝑎𝑖𝑖𝑖𝑖 + 𝑋𝑋𝑡𝑡ℎ 𝛽𝛽ℎ + 𝑋𝑋𝑡𝑡𝑙𝑙 𝛽𝛽𝑙𝑙 + ∑𝑡𝑡 𝛾𝛾𝑡𝑡 1[𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦 = 𝑡𝑡] + 𝜀𝜀𝑖𝑖ℎ𝑙𝑙𝑙𝑙 ,
where 𝑆𝑆𝑖𝑖ℎ𝑙𝑙𝑙𝑙 is an indicator variable of school attendance of girl i whose age is between 6 and 25
in hth household in a village l in year t , which takes 1 if she attends school and zero otherwise;
𝑎𝑎𝑎𝑎𝑎𝑎𝑖𝑖𝑖𝑖 is age dummies; 𝑋𝑋𝑡𝑡ℎ is a set of household characteristics including years of schooling of
mother and father, the number of sibling(s), and total amount of owned land; 𝑋𝑋𝑡𝑡𝑙𝑙 is a set of
village characteristics including the distance from Dhaka to a village, distance from the upazila
headquarter to a village, distance from the district headquarter to a village, an indicator variable
of electrification in a village—which takes one if it has access to electricity and zero otherwise
(access to electricity), the percentage of non-farm labor force in a village, the percentage of
household with migrants in a village, and the percentage of non-farm labor force in a village at
10
baseline (1988); 1[.] is an indicator function which takes one if the statement inside the bracket is
true, and zero otherwise; and 𝜀𝜀𝑖𝑖ℎ𝑙𝑙𝑙𝑙 is an error term. Our variable of interest, the proportion of
non-farm village labor force, is constructed in the following way by using the same household
panel data: total non-farm village labor force is divided by total village labor force. Labor force
in this context means those who take either farming occupation, non-farming occupation,
Second, we examine female labor force participation decisions with special attention to
the impact of accumulated human capital. We adopt the following multinomial logit model:
𝑒𝑒𝑒𝑒𝑒𝑒(𝑎𝑎1 𝑎𝑎𝑎𝑎𝑎𝑎𝑖𝑖𝑖𝑖 +𝑏𝑏1 𝐻𝐻𝑖𝑖𝑖𝑖 +𝑏𝑏+∑𝑡𝑡 𝑐𝑐𝑡𝑡 1[𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦=𝑡𝑡])
(4) Pr(𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝑖𝑖ℎ𝑙𝑙𝑙𝑙 = 𝑗𝑗) = ∑𝑘𝑘=6 𝑒𝑒𝑒𝑒𝑒𝑒(𝑎𝑎1 𝑎𝑎𝑎𝑎𝑎𝑎𝑖𝑖𝑖𝑖 +𝑏𝑏1 𝐻𝐻𝑖𝑖𝑡𝑡 +𝑏𝑏+∑𝑡𝑡 𝑐𝑐𝑡𝑡 1[𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦=𝑡𝑡])
, j = 0, … 6,
where 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝑖𝑖ℎ𝑙𝑙𝑙𝑙 is an indicator variable denoting the choice of a women i in hth household in a
village l in year t with respect to jth occupation. The variable 𝐻𝐻𝑖𝑖𝑖𝑖 is the years of schooling of
(5) 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑖𝑖ℎ𝑙𝑙𝑙𝑙 = 𝑓𝑓0 + 𝑓𝑓1 𝑎𝑎𝑎𝑎𝑎𝑎𝑖𝑖𝑖𝑖 + 𝑔𝑔1 𝐻𝐻𝑖𝑖𝑖𝑖 + 𝑋𝑋𝑡𝑡𝑙𝑙 𝑔𝑔𝑙𝑙 + ∑𝑡𝑡 ℎ𝑡𝑡 1[𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦 = 𝑡𝑡] + 𝑖𝑖𝑖𝑖ℎ𝑙𝑙𝑙𝑙 ,
We explore fertility decisions by estimating a linear model of the number of children as follows:
(6) 𝐶𝐶ℎ𝑖𝑖𝑖𝑖𝑖𝑖0_12𝑖𝑖ℎ𝑙𝑙𝑙𝑙 = 𝑘𝑘0 + 𝑘𝑘𝑎𝑎𝑎𝑎𝑎𝑎𝑖𝑖𝑡𝑡 + 𝑙𝑙1 𝐻𝐻𝑖𝑖𝑖𝑖 + 𝑋𝑋𝑡𝑡𝑙𝑙 𝑙𝑙𝑙𝑙 + ∑𝑡𝑡 𝑚𝑚𝑡𝑡 1[𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦 = 𝑡𝑡] + 𝑛𝑛𝑖𝑖ℎ𝑙𝑙𝑙𝑙 ,
where 𝐶𝐶ℎ𝑖𝑖𝑖𝑖𝑖𝑖0_12𝑖𝑖ℎ𝑙𝑙𝑙𝑙 is the number of children whose age is between 0 and 12 of woman i in
11
4. Results
Estimation results of four outcome variables of women empowerment, i.e., school attendance,
labor force participation, marriage, and fertility models, are shown in Tables 7–11. In Table 7,
which reports the estimated coefficients of the school attendance regression model of equation
(3), the proportion of non-farm labor force participation at the village level has positive and
statistically significant coefficients throughout the different econometric models controlling for
potential endogeneity bias. 4 The columns (4) to (6) and (9) to (10) show instrument variable
(IV) estimates treating the proportion of the non-farm labor force, migrated family members,
and the proportion of non-farm labor force at baseline (in year 1988) in each village as
endogenous. Since non-farm labor force participation and migration decisions would be affected
identifying IVs. Specifically, we use the following variables as IVs: distance from the village to
Dhaka; distance from the village to the upazila headquarter; distance from the village to the
district headquarter; and a dummy variable for electrification which takes one if the village is
electrified.
Results also show that household characteristics such as father’s education, mother’s
education, the number of siblings, and the amount of owned land are statistically significant and
positively associated with female school enrollment. Moreover, the coefficients of the year
dummy of 2000 and 2008 (base year is 1988) are positive and statistically significant, and the
size of the coefficients becomes large from 2000 to 2008, which indicates that female school
enrollment has improved over years. In sum, the overall results support a hypothesis that
4
We also estimated models with household or dynasty fixed effects. While the statistical significance of
the non-farm labor proportion variable is weakened, qualitative results are still maintained. This is not
necessarily inconsistent because the fixed effects per se capture the effect of exposure to village-level
non-farm labor participation and thus the proportion variable captures merely time fluctuations of the
proportion.
12
village-level exposure to non-farm labor participation facilitates school attendance of female
students.
We also apply the same regression model to boys ages 6 to 25 to compare the impact of
non-farm exposure on female school enrollment with male school enrollment. The results are
shown in Appendix 2. It is notable that the proportion of the non-farm labor force at the village
level has small and statistically insignificant coefficients in most cases, which is in contrast to
the case of girls. In our study, it seems clear that the development of non-farm sectors favored
girls’ education.
To investigate whether improved female school attendance has been motivated, at least
partially, by the improved prospect for non-farm labor market participation, we follow Estudillo
et al. (2013) and explore how occupational choice has been affected by individual education
level, which we assume is positively associated with village-level exposure to non-farm labor
participation. Tables 8 and 9 show the estimation results of the multinomial logit regression
model of female occupational choice in which the choice of being a housewife is taken as the
default category. The coefficients of years of education are positive and statistically significant,
even though coefficients of other categories (farming or/and primary occupation and housewife)
are either smaller in size or statistically insignificant. Thus, it is clear that years of education,
which we assume are partially a proxy of the proportion of the non-farm labor force in each
The estimation results of female marital status by different age groups is shown in Table
10. Again, the negative and statistically significant coefficients of years of education can be
observed. It is also notable that the coefficients are larger among the younger age group and
shrink as women age (an additional one year of education reduces the probability of marriage by
3.5 percent among women aged 15–19, 1.9 percent among women aged 20–24, and 0.4 percent
among women aged 25–29). This is consistent with what was observed in the descriptive
statistics in the previous section, that the impact of non-farm employment is particularly
13
impactful in reducing early marriage. Yet, the negative coefficients on education in the marriage
Particularly, among the youngest age group of 15–19, the point estimate with household fixed
effects becomes substantially small [specification (4) in Table 10]. This suggests that education
does not directly affect delayed marriage. An alternative interpretation of estimated results in
Table 10 is that unobserved household-level progressive norms facilitate both female education
Finally, Table 11 shows the regression results of the number of children under the age of
12 per woman aged 15–35. In our assumption, women who are exposed to non-farm
occupation, delay their marriage, and therefore, have fewer children. The results we gain in
Table 11 seem to be partially consistent with our story. The coefficient of years of education is
negative and statistically significant even with the household fixed effect. An additional one year
of schooling reduces the number of children by 0.055, which is a reduction of 4.1 percent, 5.3
percent, and 7.5 percent in the years 1988, 2000, and 2008, respectively. Hence, we obtain
supportive evidence of the model described in Figure 9, which postulates that the increased
female human capital (years of education) enhances FLFP in non-farm sectors, increases the age
of female marriage, and reduces the number of children a woman would have in her life.
MFI borrowers and non-borrowers. For this purpose, we utilize the latest two waves of surveys
in 2004 and 2008 which contain information about intra-household decision making patterns.
empowerment through the attainment of bargaining power over productive opportunities in the
household.
14
The results reported in Table 12 suggest that MFI participation can stimulate female
Indeed, if a household borrows from MFIs, the probability of female decision-making increases
by 7.3 percent on vegetable productions and 8.3 percent on sewing activities. The results are
statistically significant at 5 percent in both cases. In contrast, no such impact of MFI borrowing
can be observed on male decision-making on household production activities. Thus, our results
are consistent with our hypothesis that MFI borrowing enhances women’s empowerment by
5. Concluding Remarks
survey data. First, we investigated the evolution of female labor force participation in the
non-farm sector and educational attainment. Bangladesh has achieved the rapid industrialization
of export-oriented non-farm industries such as the garment and pharmaceutical industries, which
has provided opportunities particularly for relatively educated women to work outside their
homes. From analysis of household panel data collected from 1988 to 2008, we found that
educational attainment matters for non-farm labor market participation for women and that
examined how roles of women in household decision-making have changed over the same time
periods and found that increased participation in microfinance activities is associated with a
larger role for women in making household decisions on vegetable production and sewing. The
Bangladesh is often cited as a key to improving women’s decision-making power within each
15
progressive norms. Such changing norms may be difficult to quantify by nature but not entirely
impossible. For example, it has been found in the case of Brazil that exposure to a “family soap
opera with fewer children” resulted in a decline in fertility rates among people living in high
fertility societies (La Ferrara, Chong, and Duryea 2012). This is an area for future research in
Bangladesh.
16
References
Abadian, S. 1996. “Women’s Autonomy and Its Impact on Fertility.” World Development 24 (12):
1793-1809.
Alam, S. 2012. “The Effect of Gender-Based Returns to Borrowing on Intra-Household Resource
Allocation in Rural Bangladesh.” World Development 40 (6): 1164-80.
Amin, M.N. and T. Sonobe. 2014. “The Success of the Industrial Development Policy in the
Pharmaceutical Industry in Bangladesh.” In State Building and Development, edited by K.
Otsuka and T. Shiraishi. Taylor and Francis.
Amin, S., I. Diamond, R.T. Naved, and M. Newby. 1998. “Transition to Adulthood of Female
Garment-factory Workers in Bangladesh.” Studies in Family Planning 29 (2): 185-200.
Anderson, S., and M. Eswaran. 2009. “What Determines Female Autonomy? Evidence from
Bangladesh.” Journal of Development Economics 90 (2): 179-91.
Banerjee, A., E. Duflo, R. Glennerster, and C. Kinnan. 2015. “The Miracle of Microfinance?
Evidence from a Randomized Evaluation.” American Economic Journal: Applied Economics 7
(1): 22-53.
Boserup, E. 1970. Women’s Role in Economic Development. London: Earthscan.
Duflo, E. 2012. “Women Empowerment and Economic Development.” Journal of Economic
Literature 50 (4): 1051-79.
Estudillo, J, P., T. Matsumoto, H.C.Z. Uddin, N.S. Kumanayake, and K. Otsuka. 2013. “Labor
Markets, Occupational Choice, and Rural Poverty in Selected Countries in Asia and
Sub-Saharan Africa,” Background Paper for the World Development Report. Washington, DC:
World Bank.
Estudillo, J, P., Y. Sawada, and K. Otsuka. 2009. “The Changing Determinants of Schooling
Investments: Evidence from Villages in the Philippines, 1985–89 and 2002–04.” Journal of
Development Studies 45 (3): 391-411.
Field, E., and A. Ambrus. 2008. “Early Marriage, Age of Menarche, and Female Schooling
Attainment in Bangladesh.” Journal of Political Economy 116 (5): 881-930.
Hahn, Y., Islam, A., Nuzhat, K., Smyth, R. and Yang H. Forthcoming (accepted in 2016).
“Education, Marriage and Fertility: Long-Term Evidence from a Female Stipend Program in
Bangladesh.” Economic Development and Cultural Change.
Hashemi, S, M., S. R. Schuler, and A.P. Riley. 1996. “Rural Credit Programs and Women’s
Empowerment in Bangladesh.” World Development 24 (4): 635-53.
Heath, R., and A. M. Mobarak. 2015. “Manufacturing Growth and the Lives of Bangladesh
Women.” Journal of Development Economics 115: 1-15.
Hossain, M., and A. Bayes. 2009. Rural Economy and Livelihoods Insights from Bangladesh.
Dhaka: AH Development Publishing House.
Kabeer, N. 1997. “Women, Wages and Intra-household Power Relations in Urban Bangladesh.”
Development and Change 28: 261-302.
———. 2001. “Conflict over Credit: Re-Evaluating the Empowerment Potential of Loans to
Women in Rural Bangladesh.” World Development 29 (1): 63-84.
Kabeer, N., and S. Mahmud. 2004. “Globalization, Gender and Poverty: Bangladesh Women
Workers in Export and Local Market.” Journal of International Development 16: 93-109.
Khandker, S., M. Pitt, and N. Fuwa. 2013. Subsidy to Promote Girls Secondary Education : The
Female Stipend Program in Bangladesh. enGender Impact : the World Bank's Gender Impact
Evaluation Database. Washington, DC: World Bank.
17
La Ferrara, E., A. Chong, and S. Duryea. 2012. “Soap Operas and Fertility: Evidence from Brazil.”
American Economic Journal: Applied Economics 4 (4): 1-31
Mahmud, S., N. M. Shah, and S. Becker. 2012. “Measurement of Women’s Empowerment in Rural
Bangladesh. ” World Development 40 (3): 610-19.
Mottaleb, K. A., and T. Sonobe. 2011. “An Inquiry into the Rapid Growth of the Garment Industry
in Bangladesh.” Economic Development and Cultural Change 60 (1): 67-89.
Osmani, L. N. K. 2007. “A Breakthrough in Women’s Bargaining Power: The Impact of
Microcredit.” Journal of International Development 19 (5): 695-716.
Pitt, M. M., and S. R. Khandker. 1998. “The Impact of Group-Based Credit Programs on Poor
Households in Bangladesh: Does the Gender of Participants Matter?” Journal of Political
Economy 106 (5): 958-96.
Pitt, M. M., S. R. Khandker, and J. Cartwright. 2006. “Empowering Women with Micro Finance:
Evidence from Bangladesh.” Economic Development and Cultural Change 54 (4): 791-831.
Rhee, Y. W. 1990. “The Catalyst Model of Development: Lessons from Bangladesh’s Success with
Garment Exports.” World Development 18 (2): 333-46.
Schuler, S, R., and E. Rottach. 2010. “Women's Empowerment across Generations in Bangladesh.”
Journal of Development Studies 46 (3): 379-96.
Sen, A. 1992. “Missing Women.” British Medical Journal 304 (6827): 587-88.
United Nations. 2014. “The Millennium Development Goals Report.” New York.
World Bank. 2012. “World Development Report.” Washington, DC.
18
Table 1. Descriptive Statistics
Long-run changes in key household and female variables
19
Table 2. Occupational composition by year and gender (%) : ages 15-19
20
Table 3. Occupational composition by year and gender (%) : ages 30-35
21
Table 4. Proportion of married women by year and age group
Table 5. Average number of children below the age of 12 per woman aged 15-35
Mean
(Standard Deviation)
1988 1.35
(1.62)
2000 1.04
(1.32)
2008 0.73
(1.12)
22
Table 6. Average number of children below the age of 12 per woman aged 15-35
23
Table 7. Female school attendance regression
Dependent variable: 1= if a girl aged between 6 and 25 enrolls in schooling
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
VARIABLES
Method OLS OLS OLS IV IV FE FE IV IV
Household Household
FE FE
% of village non-farm labor force + 0.0020*** 0.0018** 0.0058*** 0.0035 0.0036* 0.0039** 0.027*
(0.00073) (0.00072) (0.0020) (0.0037) (0.0019) (0.0019) (0.016)
% of households with migrants in 0.00098*** -0.0024 0.00077* -0.0069
village+
(0.00018) (0.0028) (0.00045) (0.0046)
% of village non-farm labor force in 0.0023** 0.0064***
1988+
(0.0010) (0.0023)
Access to electricity in survey year 0.031** 0.038*** 0.036** -0.035 -0.040
(0.014) (0.014) (0.017) (0.029) (0.029)
Father's education 0.0068*** 0.0066*** 0.0062** 0.0070*** 0.0076*** 0.0066*** -0.00095 -0.00050 0.00018 -0.0072
(0.0020) (0.0020) (0.0024) (0.0020) (0.0022) (0.0025) (0.0050) (0.0050) (0.0047) (0.0059)
Mother's education 0.015*** 0.014*** 0.017*** 0.014*** 0.017*** 0.015*** 0.0089 0.0085 0.00070 0.011*
(0.0026) (0.0026) (0.0033) (0.0027) (0.0056) (0.0035) (0.0059) (0.0059) (0.0062) (0.0065)
Number of sibling(s) 0.0067* 0.0070* 0.0061 0.0055 0.0061 0.0050 0.0063 0.0070 0.011* 0.0022
(0.0037) (0.0036) (0.0041) (0.0038) (0.0042) (0.0043) (0.0065) (0.0065) (0.0056) (0.0069)
Total owned land 0.033*** 0.033*** 0.035*** 0.039*** 0.033*** 0.041*** 0.018 0.018 0.0057 0.021
(0.0080) (0.0078) (0.0088) (0.0089) (0.012) (0.0092) (0.016) (0.016) (0.016) (0.017)
Year==2000 0.19*** 0.11*** 0.22*** 0.17*** 0.40 0.23*** 0.22*** 0.15*** 0.037 0.87**
(0.017) (0.024) (0.017) (0.023) (0.27) (0.017) (0.028) (0.052) (0.12) (0.41)
Year==2008 0.21*** 0.11*** 0.22*** 0.20*** 0.44 0.24*** 0.29*** 0.21*** 0.11 0.97**
(0.019) (0.027) (0.020) (0.020) (0.28) (0.017) (0.033) (0.057) (0.11) (0.45)
Constant -0.22*** -0.24*** -0.25*** -0.33*** -0.24** 0.0052 -0.24*** -0.26***
(0.038) (0.038) (0.043) (0.032) (0.11) (0.040) (0.063) (0.064)
Observations 4,442 4,442 3,344 4,321 4,321 3,254 4,442 4,442 3,954 3,954
R-squared 0.529 0.533 0.507 0.521 0.488 0.499 0.739 0.739 0.505 0.482
Household FE NO NO NO NO NO NO YES YES YES YES
Cluster standard error YES YES YES YES YES YES YES YES YES YES
F statistic 37.6 10.5 33.11 7.70 5.60
Hansen J statistic 2.76 1.78 3.61
p-value of Hansen J statistic 0.43 0.41 0.31
24
Notes: a Robust standard errors clustered at household level are shown in parentheses. b
*** p<0.01, ** p<0.05, * p<0.1. c We also include the following control
variables which are not shown in the table: age dummy variables. + Indicates endogenous variables. e We used the following variables as exclusion restrictions:
d
distance from village to Dhaka (km_dhaka_village); distance from village to upazila headquarter (km_upazilahq); distance from village to district headquarter
(km_districthq); and a dummy variable for electrification which takes one if the village is electrified.
25
Table 8. Multinomial Logit Regression Results
Dependent variable: female occupation: 1=Farming/Primary, 2=Non-farming/Non-primary, 3=Housewife, 4=Unemployed,
5=Student, 6=Others: among women aged between 15 and 35
(1) (2) (3) (4) (5)
VARIABLES Farming/Primary Non-farming Unemployed Student Others
/Non-primary
26
Table 9. Multinomial Logit Regression with Household Fixed Effect Results
Dependent variable: female occupation: 1=Farming/Primary, 2=Non-farming/Non-primary, 3=Housewife, 4=Unemployed,
5=Student, 6=Others: among women aged between 15 and 35
(1) (2) (3) (4) (5)
VARIABLES Farming/Primary Non-farming/ Unemployed Student Others
Non-primary
27
Table 10. OLS and Household Fixed Effect Regression Results
Dependent variable: 1=if a woman in below age groups is married
(1) (2) (3) (4) (5) (6)
OLS OLS OLS Village FE Village FE Village FE
VARIABLES age 15-19 age 20-24 age 25-29 age 15-19 age 20-24 age 25-29
28
Table 11. OLS and Household Fixed Effect Regression Results
Dependent variables: number of children below the age of 12 per woman aged between 15 and 35
VARIABLES OLS FE
29
Table 12. Fixed Effect Regression Results
Dependent variable: decision-making dummy on household production activities by gender
Female Male
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
VARIABLES livestock poultry vegetable fruit/tree sewing livestock poultry vegetable fruit/tree sewing
Borrow from MFI 0.00818 -0.00647 0.0731** -0.0102 0.0829** -0.00630 0.00380 0.0125 -0.0243 -0.00145
(0.0159) (0.0275) (0.0352) (0.0202) (0.0332) (0.0238) (0.00623) (0.00852) (0.0382) (0.00730)
Land asset -0.00417 -0.0374 0.0217 -0.00596 -0.00463 0.0506** -0.00942 0.0161 0.0764* -0.00606
(0.0202) (0.0455) (0.0426) (0.0194) (0.0295) (0.0250) (0.00909) (0.0101) (0.0418) (0.00595)
Livestock asset -0.0118 -0.000996 0.0620 -0.137*** 0.0640 0.00288 0.0141 0.0116 0.114 -0.00150
(0.0266) (0.0653) (0.0789) (0.0385) (0.0797) (0.0507) (0.0133) (0.0142) (0.0720) (0.00890)
Business capital asset -0.00432 0.00984 0.0119 0.00210 0.00403 -0.00540 -0.00270 0.00861 0.00200 -0.000184
(0.00362) (0.0132) (0.0144) (0.00506) (0.0277) (0.00693) (0.00276) (0.0134) (0.0143) (0.000909)
Equipment asset 0.00762 0.0912* -0.0787 0.0133 0.0503 0.0344 0.00664 0.0691** 0.0187 0.00305
(0.0170) (0.0523) (0.0698) (0.0198) (0.0634) (0.0444) (0.00967) (0.0299) (0.0623) (0.00346)
% of Female 0.0567 0.0201 0.0472 0.158** 0.0639 -0.0201 0.0135 -0.0914** -0.117 -0.0152
(0.0548) (0.100) (0.110) (0.0599) (0.0938) (0.0741) (0.0325) (0.0375) (0.0905) (0.0207)
Observations 3,451 3,451 3,451 3,451 3,383 3,451 3,451 3,451 3,451 3,383
R-squared 0.001 0.042 0.008 0.014 0.058 0.079 0.007 0.028 0.054 0.001
Number of sample 1,747 1,747 1,747 1,747 1,741 1,747 1,747 1,747 1,747 1,741
Notes:
a
Year and household fixed effects are included in all regressions.
b
MFI includes Grameen Bank.
c
Robust standard errors in parentheses (62 village clusters).
d
*** p<0.01, ** p<0.05, * p<0.1
30
Figure 1. Analytical Framework of Women Empowerment in Bangladesh
31
Figure 2. Location of sample villages
32
Figure 3. Average Years of Schooling by Gender and Age Group
0
Year 1988 Year 2000 Year 2008
33
Figure 4. Average School Enrollment Rate by Gender and Age
Female Male
1.2 1.2
1 1
0.8 0.8
0.6
0.6
0.4
0.4
0.2
0.2
0
age 7 8 9 10 11 12 13 14 15 16 17 18 0
6 age 7 8 9 10 11 12 13 14 15 16 17 18
6
Year 1988 Year 2000 Year 1988 Year 2000
Year 2008 Year 2008
34
Figure 5. Proportion of the Married by Gender and Age Group
Female Male
100 100
90 90
80 80
70 70
60 60
50 50
40 40
30 30
20 20
10 10
0 0
Year 1988 Year 2000 Year 2008 Year 1988 Year 2000 Year 2008
Age 10-14 Age 15-19 Age 20-14 Age 10-14 Age 15-19 Age 20-24
Age 25-29 Age 30-34 Age 25-29 Age 30-34
35
Figure 6. Average Number of Children below the Age of 12 per Woman by Age
Group
2.5
1.5
0.5
0
1988 2000 2008
36
Figure 7. Changes in the Decision-making on Household Production Activities
by Male, Female, and jointly from 2004 to 2008
37
Figure 8. Changes in the Decision-making on Household Production Activities
by Male, Female, and Jointly from 2004 to 2008 in Microfinance-Borrowing
Households and Non-Borrowing Households
38
Figure 9. Models of non-farm sector development and women’s empowerment
39
Appendix 1. The first-stage regression results for endogenous variables in Table 7:
(1) (2) (3) (4)
VARIABLES % of village non-farm labor % of households with % of village non-farm labor % of households with
force migrants in a village force migrants in a village
40
Appendix 2. Male school attendance regression (with Household Fixed Effect)
Dependent variable: 1= if a boy aged between 6 and 25 enrolls in schooling
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
VARIABLES
Method OLS OLS OLS IV IV IV FE FE IV-FE IV-FE
% of village non-farm labor force + 0.00075 0.00053 0.0084*** 0.014*** 0.00064 0.00067 0.015
(0.00079) (0.00078) (0.0022) (0.0041) (0.0022) (0.0022) (0.018)
% of households with migrants in village+ 0.0011*** 0.0084** 0.00013 -0.0032
(0.00024) (0.0036) (0.00051) (0.0039)
% of village non-farm labor force in 1988+ 0.0017 0.0089***
(0.0011) (0.0025)
Access to electricity in survey year 0.036** 0.042*** 0.023 -0.024 -0.025
(0.015) (0.015) (0.019) (0.031) (0.031)
Father's education 0.016*** 0.015*** 0.019*** 0.015*** 0.0087** 0.018*** 0.0020 0.0020 0.0046 0.0013
(0.0023) (0.0023) (0.0028) (0.0024) (0.0043) (0.0029) (0.0058) (0.0058) (0.0066) (0.0051)
Mother's education 0.013*** 0.013*** 0.013*** 0.0097*** 0.0027 0.0099** -0.0025 -0.0025 -0.0016 0.00086
(0.0032) (0.0032) (0.0039) (0.0034) (0.0055) (0.0040) (0.0069) (0.0069) (0.0063) (0.0059)
Number of sibling(s) -0.0066* -0.0064* -0.0077* -0.011** -0.015** -0.013** -0.0038 -0.0036 -0.0079 -0.010
(0.0039) (0.0039) (0.0046) (0.0044) (0.0059) (0.0052) (0.0084) (0.0085) (0.0073) (0.0084)
Total owned land 0.048*** 0.049*** 0.040*** 0.059*** 0.080*** 0.048*** 0.0016 0.0016 -0.0099 -0.0088
(0.0076) (0.0076) (0.0089) (0.0095) (0.016) (0.011) (0.017) (0.017) (0.017) (0.016)
Year==2000 0.11*** 0.016 0.11*** 0.055** -0.73** 0.11*** 0.18*** 0.17*** 0.042 0.44
(0.018) (0.027) (0.019) (0.025) (0.34) (0.019) (0.032) (0.053) (0.15) (0.34)
Year==2008 0.099*** -0.0030 0.093*** 0.071*** -0.75** 0.11*** 0.20*** 0.19*** 0.059 0.47
(0.021) (0.031) (0.024) (0.023) (0.35) (0.021) (0.038) (0.059) (0.14) (0.37)
Constant 0.031 0.0082 0.070 -0.25*** -0.48*** -0.23*** 0.039 0.036
(0.054) (0.053) (0.057) (0.038) (0.11) (0.042) (0.074) (0.076)
Observations 4,559 4,559 3,422 4,387 4,387 3,303 4,559 4,559 4,158 4,158
R-squared 0.369 0.373 0.356 0.351 0.143 0.345 0.661 0.661 0.349 0.356
Household FE NO NO NO NO NO NO YES YES YES YES
Cluster standard error YES YES YES YES YES YES YES YES YES YES
F statistic 33.8 8.87 29.6 6.52 6.71
Hansen J statistic 12.1 4.09 4.85
p-value of Hansen J statistic 0.0071 0.13 0.18
Notes: a Robust standard errors clustered at household level are shown in parentheses. b *** p<0.01, ** p<0.05, * p<0.1. c We also include the following control
variables which are not shown in the table: age dummy variables. d + indicates endogenous variables. e We used the following variables as exclusion restrictions:
distance from village to Dhaka (km_dhaka_village); distance from village to upazila headquarter (km_upazilahq); distance from village to district headquarter
(km_districthq); and a dummy variable for electrification which takes one if the village is electrified.
41
Abstract(In Japanese)
要約
バングラデシュの農村における、女性のエンパワーメントの変化、およびそれ
らを促す要因を解明することが本研究の目的である。バングラデシュは 1980
年代から経済が急速な産業化を経験すると同時に、NGO などのアクターにより
マイクロファイナンスが全国的に広く普及した。本研究ではそれらの背景を踏
まえ、非農業セクターの発展とマイクロファイナンスの普及が、どのように女
性の教育、労働参加、家計における意思決定能力などの、女性のエンパワーメ
ント指標に貢献してきたかを推計している。1988-2008 年の農村パネルデータ
を使用した分析結果では、村における非農業労働者の比率が女性の教育と労働
参加とに正の関係性がある一方、若年層における既婚率と出生率とに負の関係
性があることが示された。またマイクロファイナンスを採択している家計では、
家計内の経済活動における女性の意思決定能力が高いことが明らかとなった。
42
Working Papers from the same research project