Sei sulla pagina 1di 30

Original Article

Labour Market Outcomes in Bangladesh: The Role of Poverty and


Gender Norms
Sarah Bridgesa,*, David Lawsonb and Sharifa Begumc
a
University of Nottingham, Nottingham.
b
University of Manchester, Manchester.
c
Bangladesh Institute of Development Studies (BIDS), Dhaka.
*E-mail: sarah.bridges@nottingham.ac.uk

Abstract It is frequently argued that female participation in the labour market is important for
economic growth and poverty reduction. Despite this, the role that extreme poverty and gender
norms play in influencing such participation is far from fully understood. This article uses nationally
representative household data to address this issue using data for Bangladesh. In line with prior –
largely qualitative – work, we find that extreme poverty is indeed important in explaining some of
the growth in female employment in Bangladesh; we find evidence of a positive relationship between
extreme poverty and participation, especially for women. In addition, we find that among young
single women there appears to be a growing acceptance of their employment in the labour market.
Despite this, rigid social and cultural norms are still apparent among other groups of women,
especially in the rural areas.

On affirme souvent que la participation des femmes au marché du travail joue un rôle important
pour la croissance économique et la réduction de la pauvreté. Pourtant, l0 influence de la pauvreté
extrême et des normes de genre sur cette participation est loin d0 être parfaitement comprise. Cet
article tente d0 apporter des éléments de réponse à cette question en s0 appuyant sur des données
représentatives sur les ménages du Bangladesh. Tout comme les travaux, essentiellement qualitatifs,
effectués précédemment, nous constatons que la pauvreté extrême est un facteur important
expliquant une partie de la croissance de l0 emploi des femmes au Bangladesh; Nous démontrons
que l0 extrême pauvreté est positivement corrélée avec cette participation, particulièrement pour
les femmes. En outre, nous constatons que les jeunes femmes célibataires semblent de plus en plus
nombreuses à accepter leur propre participation au marché du travail. Cependant, des normes
sociales et culturelles rigides continuent de régir les autres catégories de femmes, notamment en
milieu rural.

European Journal of Development Research (2011) 23, 459–487. doi:10.1057/ejdr.2011.14;


published online 5 May 2011
Keywords: economics; labour; gender; empowerment; poverty

Introduction

A common finding in many developing countries over the past few decades is that there
has been substantial growth in the participation of women in the labour market (Standing,
1999). Despite this, female employment still lags behind when compared with men in most
developing countries, and many development practitioners continue to suggest that it is
this under-utilisation of female labour, especially in formal employment, that is acting as a
constraint against further economic growth and poverty reduction in many developing
countries (Blackden et al, 2006).1 This presents a potentially interesting hypothesis in
regions such as South Asia, where, although female employment has increased rapidly

r 2011 European Association of Development Research and Training Institutes 0957-8811


European Journal of Development Research Vol. 23, 3, 459–487
www.palgrave-journals.com/ejdr/
Bridges et al

since the mid-1980s, it still remains an untapped resource throughout most of the region
(ILO, 2008),2 and at the same time large proportions of the population remain trapped in
extreme poverty.
Bangladesh provides a particularly interesting case study in this regard, with labour
force participation rates having increased steadily since the early 1980s (see Table A1),
and where robust economic growth has been accompanied by a sharp fall in poverty
(from about 58 per cent in 1991 to 40 per cent in 2005).3 However, despite this,
between one-quarter and a third of households are still in extreme poverty (Yoshida
et al, 2007; Serajuddin et al, 2010)4 and female employment rates constitute about a third
of that of male employment rates. It is also increasingly being recognized that female
participation in wage and salaried employment is particularly low, even when compared
with the rest of South Asia (ILO, 2004).
Furthermore, thoughts from global studies such as Standing (1999) and Cagatay and
Ozler (1995) suggest that, although neo-liberal policies have generally led to greater
feminization of the labour market, they may also have had adverse effects on women; that
is, on the demand side, firms may prefer to employ women simply because they represent a
compliant pool of cheap labour – an issue that is often interlinked with the fact that the
relatively few women who participate in the labour market often remain confined to the
ranks of the so-called ‘vulnerable’ employed.5
However, and contrasting some of the potential downsides, what has been established is
that for the poorest households, female employment clearly represents an important
source of income, which can assist a household’s strategy for moving out of extreme
poverty (Bangladesh CPR, 2005), in addition to increasing a woman’s autonomy both in
her home and the wider community.
Despite this, the determinants of female labour supply still remains a relatively under-
researched area among development practitioners; in particular, the interrelationship
between different labour market outcomes and poverty is still far from fully understood –
especially in relation to the factors that underpin entry to different employment outcomes
and how these determinants vary by gender. The need to focus on such an area, parti-
cularly from an econometric perspective, continues to be highlighted even in the most
recent literature for Bangladesh (see Serajuddin et al, 2010).
With this in mind, this article uses data from a large representative household survey
for Bangladesh to address these issues and identify some of the characteristics asso-
ciated with female participation. In doing so, we investigate the extent to which
gender norms dictate employment patterns and perhaps more importantly, in light of the
above discussion, the extent to which they are acting as a constraint against greater
participation.6
Finally, we look more closely at some of the reasons as to why the women who par-
ticipate often remained confined to employment in the less productive sectors of the
economy. As already mentioned, for many women it is not just employment per se that is
important, rather the nature of this employment is also important, as it can strongly affect
their economic freedom and self-determination.7 Anderson and Eswaran (2009), using
data for rural Bangladesh, find that wives working on their husbands’ farms often have no
more autonomy than those who are just housewives, whereas those who earn independent
income have considerably more autonomy.8
We also broaden the analysis of labour supply for Bangladesh to include that of men
who are often neglected in studies of this kind. However, given the importance of employ-
ment as a source of income for the poor, we argue that it is vital that all working-age

460 r 2011 European Association of Development Research and Training Institutes 0957-8811
European Journal of Development Research Vol. 23, 3, 459–487
Labour Market Outcomes

individuals, not just women, participate in the labour market in sustainable jobs, and in
doing so make the following key points.
We find that poverty is indeed important in explaining participation decisions,
especially for women. Such a finding persists after controlling for socio-demographic and
human capital variables. However, there also seems to be a segmentation of the labour
market along poverty lines in that certain employment outcomes seem to be the preserve
of the richer households.
We also find that among young single women there appears to be a growing acceptance
of their employment in the labour market. Such a finding persists after interacting marital
status with extreme poverty and is thus not completely down to economic necessity,
although economic necessity may have provided the catalyst.
However, although attitudes towards women and work are clearly changing, there is
still a long way to go. The effect of cultural norms on employment patterns is still readily
apparent: women who are married, and/or are living in households with dependants are
significantly less likely to participate in the labour market. This highlights the ‘double
burden’ of having to combine domestic activities with work-related activities, which
women often face when they attempt to enter the labour market. In contrast, for men,
being married is associated with their greater involvement in the labour market, given their
role as the ‘traditional’ breadwinner. There is also clear evidence, especially in the rural
areas, of more conservative attitudes towards female participation, and the role of women
more generally.
The remainder of this article is outlined as follows. The next section provides a brief
background analysis of Bangladesh, whereas the subsequent section provides a description
of the data, and a summary of the descriptive statistics. In the following section, we outline
the modelling framework. The main empirical results are outlined in the penultimate
section, whereas the last section 6 concludes the article.

Background

Although over the years a large literature has developed looking at some of the issues
surrounding female labour supply in Bangladesh, most of this research is concentrated on
urban Bangladesh and focuses on a specific city (usually Dhaka), industry (usually the
garment industry) and is based on responses from qualitative data.9 Research that focuses
on the often poorer rural regions or on the ‘choice’ workers make between different
employment outcomes (that is, self-employment versus wage employment) is rare.10 In
addition, although there seems to be a general consensus in Bangladesh (as in other
countries) that the participation of women in the labour market is determined by an
interplay of cultural, economic and labour market factors, there is still no consensus as to
the relative importance of these factors.
Bangladesh is a highly patriarchal society (as are many countries in the region) with
gender being a key factor in defining social roles, responsibilities and power relation-
ships within the family and workplace. The traditional view that men should be the main
‘provider’ still holds in many parts of the country, especially in the rural areas, with a
greater weight being attached to a woman’s domestic rather than market activities.
Women thus not only tend to rely on men for their social status and recognition, but are
also dependent on them for their economic security (Salway et al, 2003) and protection
(Kabeer, 1997). In some Bangladeshi households, these rigid cultural norms are

r 2011 European Association of Development Research and Training Institutes 0957-8811 461
European Journal of Development Research Vol. 23, 3, 459–487
Bridges et al

exacerbated further by the practice of ‘purdah’ or female seclusion, which applies


restrictions to a woman’s mobility often excluding her from public spaces and confining
her to work at home (Amin, 1997; Kabeer, 2000).11 It is clear that these patriarchal
structures create gender asymmetries in endowments, risks and constraints, which penalize
autonomous behaviour (Kabeer, 1997), and impact on a woman’s ability to participate
fully in the labour market.
Anderson and Eswaran (2009, p. 80), using data from rural Bangladesh, find that
‘ y 93 per cent of women have never been to the local bazaar, 92 per cent have never been
to the local mosque, and 68 per cent leave their residential compound at most once a
week’. In addition, Amin (1997) finds that ‘purdah’ remains a dominant influence in many
women’s lives, even in the face of extreme poverty; and qualitative work from Salway et al
(2003) finds that even among the slum dwellers of Dhaka a woman’s employment often
represents a significant loss of prestige for her entire family, and is considered to be a sign
of poverty or downward mobility. Such a view has traditionally not just been expressed by
men, but women also associate female employment outside the home with a loss of status
and honour. However, as will be seen, more recently the creation of factory-based em-
ployment has, to some extent, helped transform the parameters within which women make
their domestic and work choices.
Indeed, since the mid-1980s, there has been a rapid growth in the employment of
women in Bangladesh; female labour force participation rates have tripled since 1983. A
large part of this growth in employment is a result of the export-led expansion of the
ready-made garment (RMG) industry in the urban centres of Dhaka and Chittagong in
the 1980s and 1990s, which has generated considerable female employment (see, Afsa
(2003) for more details). Such a finding is common among other countries that have
pursued an export-led industrialization strategy based on ‘light’ industries, such as textiles,
garments and electronics (Standing, 1999). Although the demand-side factors surrounding
the female intensity of the garment industry in Bangladesh are generally widely accepted
and understood, women are thought to make a compliant pool of cheap labour in a society
that grants them few alternative labour market choices, the ‘push factors’ surrounding
their greater employment in the urban labour market appear to be less clear cut.
One frequently cited explanation for the rise in female employment in the RMG
industry, and in the urban areas more generally, is that it has arisen out of economic
necessity. In other words, it is only under extreme circumstances when households can no
longer afford the luxury of ‘purdah’ that women are able to enter the labour market.
Mahmud (1997), for example, examines female labour market behaviour in metropolitan
Dhaka and finds that participation is highest in the poorest households and declines as
household income rises. Similarly, Salway et al (2003) find that even among the slum
population of Dhaka (the majority of whom could be considered ‘poor’), women from the
better-off households are less likely to work. In addition, Mahmud (1992) finds that two
of the main characteristics associated with wage-seeking women are: being part of a
male-headed household on a low income; and being a household head.12
However, qualitative-based evidence from the likes of Kibria (1998) suggests that these
poverty-based explanations, although important, do not explain the whole story. She finds
that for women from poor urban households garment work is a way of gaining ‘economic
and social independence’, while for those who migrate from the rural areas, employment
in the garment industry is often triggered by ‘family conflicts, marital breakdown, pro-
blems of harassment and marital uncertainty’. Similarly, Paul-Majumder and Begum
(2000) argue that female garment workers do not, on the whole, come from the poorest

462 r 2011 European Association of Development Research and Training Institutes 0957-8811
European Journal of Development Research Vol. 23, 3, 459–487
Labour Market Outcomes

households, whereas Kabeer and Mahmud (2004a) find that workers from the garment
factories that operate within the country’s export-processing zones come from the more
prosperous households.13 Similarly, Khandker (1987) finds that in rural Bangladesh a
woman’s allocation of time is only partly determined by economic constraints at the
individual household level.
Thus, although poverty is undoubtedly important in explaining the rise in female
participation in the urban regions (usually Dhaka), it is clearly not the only factor at work
here. In particular, it does not explain what has caused women from the so-called better-
off households where ‘purdah’ should be strongest to enter the labour market. In addition,
the role of poverty outside Dhaka is generally less understood.
Another potential drawback of much of the current work in this area is that a large
proportion of its findings are based on the results from a series of cross-tabulations
(see, for example, Amin, 1997; Kibria, 1998; Kabeer and Mahmud, 2004a). Although
insightful, these papers do not take account of the role that other variables may play in
explaining participation, and which may be highly correlated with household income and
poverty.
In particular, the role of education is often neglected. However, for women, access to
education is not only seen as an investment in their human capital, but also has social
dimensions and is considered a source of empowerment. In addition, the potential link
between education and employment is likely to be particularly pertinent for Bangladesh,
which has experienced rapid progress in expanding basic and elementary education (in-
cluding adult literacy rates) over the past two decades for the poor and for girls (see,
Hossain, 2004). In addition, there has also been a shift in attitudes at the household level
towards female education. Blunch and Das (2007), for example, find that there has been a
shift in attitudes among Bangladeshi women towards giving girls equal or better education
than boys. We investigate these issues in more detail in what follows.

Data and Descriptive Analysis

Household Income and Expenditure Survey

This article uses nationally representative household data, and rurally based community
data from the 2000 Household Income and Expenditure Survey (HIES, 2000) for Ban-
gladesh, which collects information on household expenditure, consumption and key so-
cio-demographic variables, including labour market status and sector of employment.
As the primary aim of this article is to examine the determinants of different types of
labour market outcomes, we restrict our sample to working-age adults (15 to 64 years),
providing a sample population of 20 843 in total, of which 10 463 are men and 10 380
women.
The HIES (2000) is a particularly rich data set for looking at labour market partici-
pation decisions and employment outcomes.14 Respondents are asked detailed questions
about their usual economic activity and wage employment over the past 12 months, which
is categorized into self-employment (which we classify as agricultural or non-agricultural),
and wage employment. Those in wage employment are then asked whether they are paid a
daily wage or a salaried wage.
We also make use of the poverty line estimates that have already been calculated for
this data set by the Bangladesh Bureau of Statistics and the World Bank,15 which enables

r 2011 European Association of Development Research and Training Institutes 0957-8811 463
European Journal of Development Research Vol. 23, 3, 459–487
Bridges et al

households to be classified into four poverty states ranging from ‘extreme poverty’ to
‘non-poor’. Prior empirical work has been undertaken to test the robustness of this
measure. Notably, Begum and Sen (2005) cross-reference this poverty measure and find it
to be consistent with other welfare measures such as asset indexes, and an assessment of
household poverty based on self-categorization in the Bangladesh Demographic and
Health Survey for the same year.

Descriptive Statistics

We start our analysis with a brief outline of the main characteristics of our sample. Table 1
outlines some simple descriptive statistics of the main variables used in this article.
Our average respondent is in his/her early 30s, is married, is a Muslim and lives in a
household with at least two children (with around one child under the age of 5 years).
Around half the sample has no education, which is greater for women (57 per cent)
than men (43 per cent). Similarly, a smaller proportion of women (5 per cent) than men
(13 per cent) have a post-secondary education. Turning to our measures of poverty, we
find that approximately 39 per cent of respondents are from households in the bottom
two poverty groups and we see a clear division of labour between men and women, and
participation rates are substantially higher for men (81 per cent) than women (12 per cent).
In addition, we find that those who participate are more likely to be engaged in some form
of wage employment. For women, this takes the form of salaried work, whereas for men it
takes the form of daily wage employment. In addition, judging from the rural and urban
identifiers, there appear to be significant geographical differences as to where respondents
(and hence our workers) reside, with a high proportion of our sample living in or around
Dhaka.
Next, we examine the relationship between poverty and female participation (and also
sector of employment). As with the studies outlined in the previous section, we find that
individuals in households that are poor are also more likely to participate in the labour
market, although this effect is more pronounced for men than women (Table 2).
Disaggregating this result further to look at the relationship between poverty status
and sector of employment, we find that for both men and women extreme poverty
seems to be associated with their participation in daily wage employment, whereas in the
richer households their participation tends to take the form of self-employment or salaried
employment. It should, however, be noted that for the majority of the women in
our sample, poverty is still overwhelmingly associated with their non-participation in the
labour market.16
Next, we look more closely at the relationship between employment outcomes and key
household characteristics (Table 3) – factors that are considered to be important from a
gender perspective in determining access to employment.
Here we find that important household differences emerge in this setting. In particular,
the results highlight the importance of having working-age adults in the household in
order to facilitate participation in salaried employment. This is particularly the case for
women when there are a high proportion of adult women in the household. For example,
salaried women have on average 2.1 adult women in the household, compared with 1.5
for women in agricultural self-employment. This is also combined with the presence of a
smaller number of young children (less than 5 years) in these households. Such beneficial
demographic dividends clearly enable women to enter the formal labour market, a finding
that comes through in the multinomial regression results that follow.

464 r 2011 European Association of Development Research and Training Institutes 0957-8811
European Journal of Development Research Vol. 23, 3, 459–487
Table 1: Summary statistics

Variable All Women Men

Mean Std. dev. Mean Std. dev. Mean Std. dev.


=1 if participate in any form of employment, 0 otherwise 0.47 0.50 0.12 0.32 0.81 0.39
=1 non-agricultural self-employed, 0 otherwise 0.10 0.30 0.01 0.11 0.19 0.39
=1 if agricultural self-employed, 0 otherwise 0.09 0.29 0.02 0.14 0.16 0.37
=1 if paid a daily wage, 0 otherwise 0.16 0.37 0.04 0.19 0.28 0.45
=1 if paid a salary, 0 otherwise 0.11 0.32 0.05 0.21 0.18 0.38

Household poverty status:


=1 if extreme poor, 0 otherwise 0.14 0.35 0.15 0.36 0.13 0.33
=1 if moderately poor, 0 otherwise 0.25 0.43 0.26 0.44 0.24 0.43
=1 if mid-non-poor, 0 otherwise 0.35 0.48 0.34 0.47 0.36 0.48
=1 if top non-poor, 0 otherwise 0.26 0.44 0.25 0.44 0.27 0.45

Personal characteristics:

Age of respondent (years) 32.71 12.81 32.45 12.49 32.98 13.12


Marital status
=1 if married, 0 otherwise 0.70 0.46 0.76 0.43 0.64 0.48
=1 if single, 0 otherwise 0.25 0.43 0.14 0.35 0.35 0.48
=1 if divorced/separated, 0 otherwise 0.01 0.11 0.02 0.14 0.005 0.07
=1 if widowed, 0 otherwise 0.04 0.20 0.08 0.27 0.005 0.05
=1 if sick, 0 otherwise 0.19 0.39 0.20 0.40 0.17 0.38

European Journal of Development Research Vol. 23, 3, 459–487


Human capital variables:
Highest level of educational achievement:
=1 if no education, 0 otherwise 0.49 0.50 0.57 0.49 0.43 0.49
=1 if primary incomplete, 0 otherwise 0.11 0.31 0.11 0.31 0.11 0.31

r 2011 European Association of Development Research and Training Institutes 0957-8811


=1 if primary complete, 0 otherwise 0.04 0.20 0.04 0.20 0.04 0.21
=1 if secondary incomplete, 0 otherwise 0.19 0.39 0.17 0.3 0.20 0.40
Labour Market Outcomes

=1 if secondary complete, 0 otherwise 0.08 0.26 0.06 0.23 0.09 0.29


=1 if high secondary, 0 otherwise 0.09 0.28 0.05 0.21 0.13 0.33

465
466
Table 1 continued

Variable All Women Men

Mean Std. dev. Mean Std. dev. Mean Std. dev.


Household characteristics:
Age of head of household (years) 46.21 12.93 46.12 13.39 46.30 12.46
Bridges et al

=1 if female head of household, 0 otherwise 0.05 0.23 0.07 0.26 0.04 0.19
Number of working-age adults in the household 3.54 1.73 3.41 1.71 3.67 1.74
Number of 0–5-year olds in the household 0.74 0.88 0.76 0.88 0.72 0.87
Number of 6–14-year olds in the household 1.35 1.25 1.35 1.25 1.36 1.24
Number of adults above the age of 60 years in the household 0.38 0.60 0.40 0.61 0.36 0.58
=1 if Muslim, 0 otherwise 0.90 0.30 0.91 0.29 0.90 0.30

Regional dummies (base Chittagong):


=1 if live in rural Chittagong, 0 otherwise 0.17 0.37 0.16 0.37 0.17 0.37
=1 if live in urban Chittagong, 0 otherwise 0.07 0.26 0.07 0.26 0.08 0.27
=1 if live in rural Dhaka, 0 otherwise 0.18 0.38 0.18 0.38 0.18 0.38
=1 if live in urban Dhaka, 0 otherwise 0.13 0.34 0.13 0.33 0.13 0.34
=1 if live in rural Khulna, 0 otherwise 0.08 0.27 0.08 0.27 0.08 0.27
=1 if live in urban Khulna, 0 otherwise 0.05 0.21 0.05 0.21 0.05 0.21
=1 if live in rural Rajshahi, 0 otherwise 0.17 0.37 0.17 0.38 0.16 0.37
=1 if live in urban Rajshahi, 0 otherwise 0.06 0.24 0.06 0.23 0.06 0.24
=1 if live in rural Barisal, 0 otherwise 0.07 0.26 0.07 0.26 0.07 0.26
=1 if live in urban Barisal, 0 otherwise 0.03 0.16 0.03 0.16 0.03 0.17

European Journal of Development Research Vol. 23, 3, 459–487


r 2011 European Association of Development Research and Training Institutes 0957-8811
Table 2: Percentage of individuals in each poverty category disaggregated by employment sector and gender

Women Men

Extreme Moderately Mid-non- Top Extreme Moderately Mid- Top


poor poor poor non-poor poor poor non-poor non-poor
Participation 17.40 12.11 9.57 11.04 91.07 88.56 79.38 72.77

Sector of employment:
Non-agricultural self-employment 1.81 1.33 1.05 0.91 12.5 16.61 19.33 24.33
Agricultural self-employment 1.41 1.94 1.85 2.34 10.86 17.87 17.27 15.06
Daily wage 10.61 4.85 2.34 1.06 58.93 42.09 23.28 8.46
Salary 3.49 3.99 4.33 6.73 8.78 11.99 19.5 24.92

European Journal of Development Research Vol. 23, 3, 459–487


r 2011 European Association of Development Research and Training Institutes 0957-8811
Labour Market Outcomes

467
468
Bridges et al

Table 3: Socio-economic characteristics disaggregated by sector of employment, gender and area of residence

Women Men

Non-agricultural Agricultural Daily Salary Non-agricultural Agricultural Daily Salary


self-employment self-employment self-employment self-employment

=1 if female head of household, 0 otherwise 0.36 0.39 0.49 0.22 0.02 0.01 0.04 0.05
Number of adult women in household 1.64 1.54 0.82 2.13 1.83 1.84 1.49 1.77
Number of adult men in household 1.19 1.15 1.07 1.51 2.18 2.27 1.73 2.30
Number of 0–5-year olds in the household 0.64 0.72 0.88 0.50 0.80 0.79 0.78 0.64
Number of 6–14-year olds in the household 1.25 1.31 1.21 1.20 1.39 1.46 1.34 1.21
Number of adults above the age of 60 years in the household 0.17 0.20 0.43 0.19 0.23 0.34 0.17 0.23
Number of children and 1 adult woman in HH 1.90 1.96 1.49 1.62 2.14 2.13 2.11 1.84
Number of children and between 2 and 3 adult women in HH 1.86 1.98 1.59 1.72 2.09 2.22 2.08 1.82
Number of children and more than 3 adult women in HH 2.00 3.86 1.46 1.84 3.46 3.71 3.12 2.60
At least 1 man in the household in participation 0.75 0.65 0.48 0.82 1.00 1.00 1.00 1.00
At least 1 man in household in non-agricultural self- 0.33 0.19 0.39 0.27 1.00 0.09 0.06 0.12
employment
At least 1 man in the household in agricultural self-employment 0.04 0.08 0.27 0.09 0.09 1.00 0.09 0.13
At least 1 man in the household in daily wage employment 0.27 0.32 0.47 0.22 0.07 0.11 1.00 0.10

European Journal of Development Research Vol. 23, 3, 459–487


At least 1 man in the household in salaried employment 0.20 0.14 0.34 0.43 0.12 0.12 0.06 1.00

r 2011 European Association of Development Research and Training Institutes 0957-8811


Labour Market Outcomes

At the same time, we find that households with a higher than average number of adult
women, not surprisingly, also have a higher than average number of children. Interest-
ingly, we find that men in these households are more likely to participate in flexible forms
of employment, particularly agricultural self-employment, and are less likely to be in
salaried employment. Such a finding could be reflective of changing roles within the home,
with men, under certain circumstances, being more willing to take on a greater role within
the household.
We also find that women are more likely to participate in the labour market in house-
holds where the men are also employed. There is a particularly high correlation between
male participation in daily/salaried employment and female participation in the same type of
employment. This may arise, at least, partly because of economic necessity, or may reflect
the fact that men in work are more likely to marry/live with women who also work. How-
ever, it could again be reflective of the idea that traditional gender norms are slowly begin-
ning to change within Bangladesh, making it increasingly more acceptable for women to work.
The analysis so far not only highlights that some interesting household factors may
influence employment possibilities, but also clearly provides some support for the findings
outlined in the previous section that in the face of extreme hardship ‘purdah’ or female
seclusion becomes a ‘luxury’ that households can no longer afford. We move now to
examine whether poverty continues to ‘predict’ participation and sector of employment in
a multivariate setting. In doing so, we try to identify some of the other key characteristics
aside from household characteristics and economic necessity that are associated with
employment in Bangladesh.17

Empirical Methodology

Participation Equation

We begin our multivariate analysis by estimating the probability that individual i parti-
cipates in the labour market (yi0 ¼ 1) using a standard latent variable model of the form:
yi0 ¼ b0 0 Xi0 þ go povi þ ui0 ð1Þ

where y*i0 is the latent (unobserved) propensity of participation, Xi0 are the set of
exogenous regressors thought to affect participation, including region-specific effects, povi0
is a measure of poverty status and ui0 is a random error that has a standard logistic
distribution.

Sector of Employment

Next, we examine the extent to which the predictors of participation vary with sector
of employment and in doing so allow for five distinct employment outcomes,
m: non-participation (m ¼ 0), self-employment (agricultural) (m ¼ 1), self-employment
(non-agricultural) (m ¼ 2), daily wage (m ¼ 3) and salaried employment (m ¼ 4). In the
light of the number of discrete choices involved, we model employment outcomes for
individual i using a multinomial logit model, which we motivate using a latent variable
framework of the form:

yim ¼ bm Xi þ gm povi þ uim ; m¼ 0; 1; 2; 3; 4 ð2Þ

r 2011 European Association of Development Research and Training Institutes 0957-8811 469
European Journal of Development Research Vol. 23, 3, 459–487
Bridges et al

where yim * is the latent (unobserved) propensity of participation in employment sector m;,
Xi are the set of exogenous regressors thought to affect participation and (ui ¼ ui0, ui1,
ui2, ui3, ui4) are the set of disturbances that are assumed to be distributed independently and
have a standard logistic distribution.18 As before the latent propensities are unobservable.
However, we observe the discrete outcomes yi, which are assumed to be related to the
underlying propensities through an observability criterion of the form:

yi ¼ m if yim ¼ max yim ð3Þ


j¼1;...;M

In all specifications, our exogenous variables include controls for personal demographic
variables (such as, age, marital status and health status), household characters (such as,
age of the household head, the number and age of children in the household, religion and
region of residence), human capital variables (a set of dummy variables for stages of
completed education) and a measure of the household’s poverty status.

Empirical Results

Table 4 illustrates the results of this exercise for both our logit and multinomial logit
regressions. Results are reported separately for women (panel A) and men (panel B). For
the standard logits, we reports marginal effects (evaluated at the means of the regressor
variables). However, as for the multinomial logits we are primarily interested in how a
change in a variable affects the propensity of being in a state relative to non-participation
(our base case), we report only coefficients.19
We begin by calculating the probability of participation for both men and women,
and find that in line with the summary statistics outlined in the section ‘Background’
large differences emerge between men and women in their likelihood of participation.
The ‘average’ man has a 90 per cent probability of participating in the labour market,
compared with just below 9 per cent for the ‘average’ women.20
However, of most interest in this context is the association between poverty status,
participation and sector of employment. For women, Table 4 confirms our positive re-
lationship between extreme poverty and participation; being in a household in extreme
poverty raises a woman’s likelihood of participation by 4 per cent, but is not significant for
the other poverty states (base case is non-poor). In contrast, for men the effects of poverty
are more acute; being in a household that is either extremely poor or moderately poor
raises their likelihood of participation in the labour market.21
However, in line with the summary statistics outlined in the section ‘Background’, we
find that the richest and poorest individuals are employed in very different types of
activities. We find that for both genders participation in daily wage employment (relative
to non-participation) appears to be the preserve of the ‘poorer’ households, whereas
self-employment appears to be the preserve of the ‘richer’ households. For women, this
takes the form of agricultural self-employment, whereas for men it takes the form of
non-agricultural self-employment. Interestingly, economic necessity (or poverty status in
general) appears to have little effect on participation in salaried employment. These
findings thus provide further support for the view that a household’s economic choices are
at least partially influenced by the economic needs of the household and are not ex-
clusively fixed by rigid societal norms, although it appears that only certain types of
occupations are open to the poorer households.

470 r 2011 European Association of Development Research and Training Institutes 0957-8811
European Journal of Development Research Vol. 23, 3, 459–487
Table 4: Maximum likelihood estimates of labour market outcomes: (a) Women; (b) Men
Variables Logistic Multinomial logistic regression (Coefficients)
Regression:
Probability of
participation

Marginal Self-employment Self-employment Daily wage Salary


effects (non-agricultural) (agricultural)
(a)
Household poverty status (base case top non-poor):
=1 if household extreme poor 0.039*** 0.751** 1.287*** 1.777*** 0.317
(0.010) (0.347) (0.286) (0.252) (0.202)
=1 if household moderately poor 0.012 0.408 0.797*** 1.099*** 0.149
(0.008) (0.308) (0.218) (0.241) (0.160)
=1 if household mid-non-poor 0.011 0.002 0.546*** 0.486** 0.149
(0.007) (0.284) (0.193) (0.239) (0.131)
Personal characteristics:
Age 0.015*** 0.169*** 0.188*** 0.267*** 0.181***
(0.002) (0.061) (0.052) (0.037) (0.033)
Age-squared 0.0002*** 0.002*** 0.002*** 0.004*** 0.003***
(0.00002) (0.001) (0.001) (0.000) (0.000)
=1 if single, 0 otherwise 0.082*** 0.656 0.099 1.342*** 0.987***
(0.011) (0.516) (0.519) (0.300) (0.187)
=1 if divorced/separated, 0 otherwise 0.147*** 1.356*** 0.930* 2.270*** 1.980***
(0.013) (0.484) (0.476) (0.231) (0.208)
=1 if widowed, 0 otherwise 0.128*** 1.670*** 1.298*** 2.133*** 1.293***
(0.010) (0.291) (0.268) (0.183) (0.202)

European Journal of Development Research Vol. 23, 3, 459–487


=1 if sick, 0 otherwise 0.010* 0.397** 0.146 0.065 0.071
(0.006) (0.202) (0.170) (0.133) (0.124)
Household characteristics:

r 2011 European Association of Development Research and Training Institutes 0957-8811


Age of head of household 0.001*** 0.029*** 0.013 0.024*** 0.005
(0.000) (0.010) (0.008) (0.006) (0.005)
Number of working-age adults in the household 0.025*** 0.450*** 0.494*** 0.520*** 0.151***
Labour Market Outcomes

(0.002) (0.082) (0.070) (0.056) (0.036)


Number of 0–5-year olds in the household 0.016*** 0.025 0.219** 0.369*** 0.364***
(0.004) (0.126) (0.094) (0.085) (0.075)

471
472
Table 4 continued

Variables Logistic Multinomial logistic regression (Coefficients)


Regression:
Bridges et al

Probability of
participation

Marginal Self-employment Self-employment Daily wage Salary


effects (non-agricultural) (agricultural)
Number of 6–14-year olds in the household 0.008*** 0.065 0.048 0.183*** 0.113**
(0.002) (0.085) (0.066) (0.054) (0.045)
Number of adults above the age of 60 years in the 0.026*** 0.673*** 0.520*** 0.469*** 0.162
household (0.006) (0.239) (0.192) (0.143) (0.105)
=1 if Muslim, 0 otherwise 0.039*** 0.872*** 0.347 0.985*** 0.372**
(0.008) (0.253) (0.318) (0.163) (0.155)

Highest level of educational achievement


(base case no education):
=1 if primary incomplete, 0 otherwise 0.009 0.120 0.154 0.279 0.101
(0.013) (0.441) (0.361) (0.318) (0.259)
=1 if =1 if primary complete, 0 otherwise 0.029*** 0.224 0.283 1.581*** 0.297
(0.010) (0.351) (0.222) (0.370) (0.182)
=1 if secondary incomplete, 0 otherwise 0.049*** 0.179 0.184 1.038*** 0.627***
(0.009) (0.328) (0.263) (0.275) (0.173)
=1 if secondary complete, 0 otherwise 0.029** 1.151 0.524 0.950** 0.096
(0.013) (0.749) (0.486) (0.445) (0.205)

European Journal of Development Research Vol. 23, 3, 459–487


=1 if high secondary, 0 otherwise 0.027** 0.315 1.011 1.150* 0.698***
(0.012) (0.644) (0.740) (0.622) (0.187)
=1 if rural Dhaka, 0 otherwise 0.032*** 0.815** 0.990*** 0.315 1.271***
(0.009) (0.317) (0.223) (0.228) (0.202)

r 2011 European Association of Development Research and Training Institutes 0957-8811


=1 if urban Dhaka, 0 otherwise 0.034*** 0.103 0.392 0.325 0.645***
(0.009) (0.312) (0.371) (0.307) (0.133)
=1 if rural Khulna, 0 otherwise 0.059*** 0.614 0.123 0.283 1.426***
(0.013) (0.393) (0.405) (0.282) (0.293)
=1 if urban Khulna, 0 otherwise 0.020* 0.734** 1.652 0.952*** 0.065
(0.012) (0.346) (1.023) (0.283) (0.222)
=1 if rural Rajshahi, 0 otherwise 0.010 0.867*** 0.862*** 1.014*** 1.159***
(0.008) (0.334) (0.243) (0.196) (0.203)
=1 if urban Rajshahi, 0 otherwise 0.045*** 0.443 0.718** 1.388*** 0.100
(0.011) (0.359) (0.365) (0.243) (0.200)
=1 if rural Barisal, 0 otherwise 0.041*** 1.344** 0.422 0.018 1.077***
(0.013) (0.610) (0.317) (0.325) (0.290)
=1 if urban Barisal, 0 otherwise 0.011 0.260 0.357 0.428 0.068
(0.016) (0.624) (0.500) (0.434) (0.263)
Number of Observations 10380 10380
Log-likelihood 3271.344 4385.404

(b)
Household poverty status (base case top non-poor):
=1 if household extreme poor 0.021** 0.517*** 0.711*** 1.709*** 0.233
(0.009) (0.172) (0.178) (0.166) (0.177)
=1 if household moderately poor 0.021*** 0.236* 0.035 1.417*** 0.338***
(0.007) (0.125) (0.129) (0.131) (0.127)
=1 if household mid-non-poor 0.000 0.382*** 0.130 0.660*** 0.149
(0.005) (0.095) (0.103) (0.110) (0.093)

Personal characteristics:
Age 0.028*** 0.491*** 0.462*** 0.428*** 0.455***
(0.001) (0.024) (0.025) (0.024) (0.023)
Age-squared 0.0004*** 0.006*** 0.006*** 0.006*** 0.006***
(0.0002) (0.000) (0.000) (0.000) (0.000)
=1 if single, 0 otherwise 0.100*** 1.794*** 1.380*** 1.726*** 1.594***

European Journal of Development Research Vol. 23, 3, 459–487


(0.008) (0.155) (0.170) (0.162) (0.152)
=1 if divorced/separated, 0 otherwise 0.070 1.860** 0.190 1.190 1.422
(0.046) (0.929) (0.844) (0.818) (0.886)
=1 if widowed, 0 otherwise 0.106*** 1.288*** 1.390*** 2.455*** 1.966***

r 2011 European Association of Development Research and Training Institutes 0957-8811


(0.025) (0.478) (0.470) (0.540) (0.591)
=1 if sick, 0 otherwise 0.008 0.068 0.183 0.162 0.083
(0.006) (0.110) (0.114) (0.108) (0.109)
Labour Market Outcomes

473
474
Bridges et al

Table 4 continued

Variables Logistic Multinomial logistic regression (Coefficients)


Regression:
Probability of
participation

Marginal Self-employment Self-employment Daily wage Salary


effects (non-agricultural) (agricultural)

Household characteristics:
Age of head of household 0.000 0.006 0.004 0.001 0.001
(0.000) (0.004) (0.004) (0.004) (0.004)
Number of working-age adults in the household 0.010*** 0.093*** 0.102*** 0.338*** 0.157***
(0.001) (0.024) (0.026) (0.027) (0.024)
Number of 0–5-year olds in the household 0.001 0.099* 0.128** 0.179*** 0.085
(0.003) (0.051) (0.053) (0.052) (0.053)
Number of 6–14-year olds in the household 0.001 0.070** 0.089*** 0.096*** 0.035
(0.002) (0.033) (0.034) (0.033) (0.033)
Number of adults above the age of 60 years in the 0.006 0.120 0.131* 0.273*** 0.156**

European Journal of Development Research Vol. 23, 3, 459–487


household (0.004) (0.074) (0.076) (0.076) (0.073)
=1 if Muslim, 0 otherwise 0.002 0.361*** 0.291** 0.006 0.043
(0.006) (0.122) (0.138) (0.128) (0.125)

r 2011 European Association of Development Research and Training Institutes 0957-8811


Highest level of educational achievement
(base case no education):
=1 if primary incomplete, 0 otherwise 0.008 0.234 0.318 0.010 0.311
(0.011) (0.207) (0.214) (0.194) (0.223)
=1 if =1 if primary complete, 0 otherwise 0.014* 0.346** 0.435*** 0.071 0.726***
(0.008) (0.147) (0.150) (0.138) (0.151)
=1 if secondary incomplete, 0 otherwise 0.066*** 0.874*** 0.705*** 1.845*** 0.304***
(0.006) (0.111) (0.113) (0.110) (0.114)
=1 if secondary complete, 0 otherwise 0.100*** 1.449*** 1.608*** 3.213*** 0.448***
(0.008) (0.139) (0.152) (0.190) (0.136)
=1 if high secondary, 0 otherwise 0.111*** 1.943*** 2.589*** 3.842*** 0.311**
(0.008) (0.139) (0.168) (0.213) (0.132)
=1 if rural Dhaka, 0 otherwise 0.003 0.047 1.024*** 0.300** 0.318**
(0.007) (0.129) (0.132) (0.124) (0.128)
=1 if urban Dhaka, 0 otherwise 0.027*** 0.229* 1.478*** 1.216*** 0.190
(0.007) (0.126) (0.188) (0.139) (0.117)
=1 if rural Khulna, 0 otherwise 0.018** 0.355** 0.921*** 0.617*** 1.190***
(0.008) (0.167) (0.162) (0.163) (0.183)
=1 if urban Khulna, 0 otherwise 0.025** 0.194 1.382*** 0.769*** 0.612***
(0.010) (0.181) (0.309) (0.197) (0.181)
=1 if rural Rajshahi, 0 otherwise 0.004 0.252* 1.347*** 0.120 0.816***
(0.007) (0.144) (0.139) (0.133) (0.145)
=1 if urban Rajshahi, 0 otherwise 0.013 0.430*** 0.537** 0.461*** 0.599***
(0.009) (0.167) (0.227) (0.177) (0.172)
=1 if rural Barisal, 0 otherwise 0.022*** 0.269* 0.433*** 0.647*** 0.832***
(0.008) (0.162) (0.163) (0.161) (0.166)
=1 if urban Barisal, 0 otherwise 0.025** 0.130 1.205*** 0.651*** 0.632***
(0.012) (0.220) (0.350) (0.244) (0.220)
Number of Observations 10463 10463
Log-likelihood 3028.413 12436.103

European Journal of Development Research Vol. 23, 3, 459–487


r 2011 European Association of Development Research and Training Institutes 0957-8811
Labour Market Outcomes

475
Bridges et al

However, these poverty-based explanations, although important, clearly do not explain


the full story behind the growth in female employment in Bangladesh. Interestingly, the
demographic variables such as age and marital status also show strong associations
with labour force participation. Age is represented in the quadratic form (higher-order
polynomials were rejected by the appropriate likelihood ratio tests) and has an inverted
u-shaped effect on both participation and sector of employment; participation increases
with age after which it begins to fall.
Looking at marital status, we find that single women are more likely than their married
counterparts to participate in the labour market, as are those who are divorced/separated
and widowed. Indeed, that divorced/separated women are more likely to participate
in the labour market is perhaps reflective of the gender inequalities in the law relating to
divorce in Bangladesh (and other developing countries). Women in developing countries
rarely benefit financially through divorce, which increases their need to participate in the
labour market. However, these findings persist after interacting marital status with ex-
treme poverty, which implies that these women are not participating solely out of eco-
nomic necessity (see Table A2).
Therefore, the data clearly suggest that young women and single women are more likely
to participate in the labour market, which is cause for optimism.22 The multinomial
results, however, highlight that only certain types of employment appear to be available to
single women. Here being single is compatible with wage employment (whether daily wage
or salaried), but is largely incompatible with self-employment. Salway et al (2003) find a
similar result among single women in Dhaka, and argue that such a finding represents an
important departure for young, single Bangladeshi women, and is reflective of the growing
acceptability of certain forms of employment for unmarried women (in their case work in
the garment industry). Interestingly, all forms of employment appear to be open to di-
vorced/separated and widowed women.
Similarly, qualitative evidence from Kibria (1998, p. 11) finds that among young, single,
rural migrant women garment work is seen as a way to ‘ y take care of oneself financially
and build one’s own future, thus reducing the burden and responsibility of the family for
one’s upkeep and well-being’. It is also thought to enhance a woman’s marriage prospects
by demonstrating that they can generate their own income, and perhaps even contribute
towards their dowry.
Such findings cannot, however, be unambiguously interpreted as being positive in terms
of a woman’s empowerment, as discrimination against women is not just confined to the
household, but also exists among employers. As a result, wages and working conditions
for women in Bangladesh are often poor. This is particularly true in the RMG, which
employs a high proportion of young women; hours of work are often longer than else-
where in the manufacturing sector and hourly wages are low (see, Kabeer and Mahmud,
2004b). Despite this, Kabeer and Mahmud (2004b, p. 152) argue that employment in the
garments industry has had a positive effect on women enabling them to ‘ y negotiate with
dominant family members, to postpone their age of marriage and to exercise greater choice
in who they marry, to contribute to their families and thus to be perceived and valued as
earning family members’.
In contrast, for men, being married has a large positive effect on their likelihood of
participation, which is reflective of their traditional role in the household as the ‘bread-
winner’. Although work that looks at male labour force participation in Bangladesh is
rare, such a finding is in line with the results for other developing countries (see, for
example, Glick and Sahn, 1997 and Bridges and Lawson, 2009).

476 r 2011 European Association of Development Research and Training Institutes 0957-8811
European Journal of Development Research Vol. 23, 3, 459–487
Labour Market Outcomes

In line with the summary statistics outlined in the previous section, we would also
expect household structure to have a role to play in this setting. We find that for women
the greater the number of dependants in the household the lower is their likelihood of
participation. Such a finding is in line with other studies for Bangladesh (see, for example,
Salway et al, 2003), and is clearly reflective of strong gender norms that still prevail at the
household level.
Indeed, turning to the multinomial logit results, we find that, not surprisingly, the
presence of children in the household has a positive effect on a woman’s participation in
agricultural self-employment, whereas lowers her likelihood of being in wage employment.
Such a finding invariably comes down to the flexibility of self-employment relative to wage
employment, which clearly gives women with young children more opportunities to
combine childcare with employment. In a related issue insufficient and/or high costs of
childcare arrangement have been found to be a barrier against their participation in paid
work (Lokshin et al, 2004). Interestingly, for men we also find that the presence of children
in the household has a positive effect on their participation in any form of self-employ-
ment, whereas lowers their likelihood of being in daily wage employment. Such a finding
could perhaps be reflective of the increased earning opportunities that have begun to
emerge for women in the labour market, and be indicative of slight changes in the caring
role within families, although the earlier descriptive data would seem to indicate that this
effect dominates in households where there are above average number of adult women.
However, Table 4 also illustrates that marital status and the presence of children have
independent significant effects. Such findings imply that being married has a negative
effect on participation over and above the potential conflict between childcare and in-
come-earning activities that we outlined above. Salway et al (2003, p. 888) find a similar
result and argue that it is ‘ y indicative of male control over female entry into the labour
force within marriage y ’. However, this could also come down to life cycle effects.
Kabeer and Mahmud (2004b), for example, suggest that there is a strong life cycle element
attached to a woman’s participation in the labour market. They focus on the RMG
industry and argue that garment work is largely suited to young unmarried women as
hours of work are often long and inflexible.
Turning now to look at the role of human capital (here proxied for by dummy variables
for educational attainment), we find that at first sight our human capital variables appear
to have a spurious effect on participation. However, closer examination reveals that two
types of Bangladeshi women participate in the labour market; those who have no edu-
cation and those with high levels of education. That women with no education participate
is perhaps down to economic necessity. However, the fact that those with high levels of
education also participate is particularly noteworthy from a policy perspective. Although
an educated woman is not necessarily an ‘empowered’ one, education is nevertheless
considered a source of empowerment (see, for example, Kishor, 2000). In this setting, more
educated women should be better able to renegotiate rigid gender norms in order to
participate in the labour market. In addition, more educated women should also have
more labour market opportunities available to them, making it easier for them to gain
employment. Khandker (1987) finds a similar result for urban Bangladesh and argues that
the higher a woman’s educational attainment, the greater is her opportunity cost of not
producing cash income. However, as we have seen, education is not the only barrier
that women have to overcome in order to gain entry into the labour market; they also have
to overcome gender stereotypes, not only at a household level, but also by employers.
Indeed, Rahman (2005) finds that many employers still prefer to employ men for jobs

r 2011 European Association of Development Research and Training Institutes 0957-8811 477
European Journal of Development Research Vol. 23, 3, 459–487
Bridges et al

that they have traditionally performed, arguing that there is nothing to be gained from
hiring women.
Important regional differences also emerge. Notably, for women, living in urban areas
has a positive effect on participation. Such a finding is at least partly reflective of the
increased labour market opportunities that are available to women in these areas, notably
garment work. We consider this further in the following sub-section.

Urban or Rural?

Next, we examine the extent to which participation decisions are affected by any
urban/rural split in attitudes towards women. In doing so, we estimate the probability of
participation, again using a logistic regression, for women and men in urban and rural
Bangladesh, respectively (Table 5).23
We begin by calculating the probability of participation for both men and women,
and find that an ‘average’ woman has a probability of participation of approximately

Table 5: Maximum likelihood estimates of probability of participation (urban/rural) – Marginal


effects

Variables Women Men

Rural Urban Rural Urban


Household poverty status (base case top non-poor):
=1 if household extreme poor 0.028*** 0.066*** 0.035 0.015*
(0.009) (0.023) (0.023) (0.009)
=1 if household moderately poor 0.007 0.022 0.062*** 0.010
(0.008) (0.019) (0.017) (0.007)
=1 if household mid-non-poor 0.001 0.043*** 0.019* 0.006
(0.008) (0.015) (0.011) (0.005)

Personal characteristics:
=1 if single, 0 otherwise 0.052*** 0.121*** 0.153*** 0.077***
(0.012) (0.022) (0.017) (0.009)
=1 if divorced/separated, 0.103*** 0.212*** 0.115 0.048
0 otherwise (0.013) (0.027) (0.088) (0.056)
=1 if widowed, 0 otherwise 0.102*** 0.157*** 0.025 0.115***
(0.010) (0.022) (0.100) (0.023)

Household characteristics:
Number of 0–5-year olds in the 0.010*** 0.023*** 0.013* 0.002
household (0.003) (0.008) (0.007) (0.003)
Number of 6–14-year olds in the 0.005** 0.010* 0.002 0.001
household (0.002) (0.005) (0.004) (0.002)
Number of adults above the age 0.029*** 0.013 0.004 0.007*
of 60 years in the household (0.006) (0.012) (0.008) (0.004)
=1 if Muslim, 0 otherwise 0.041*** 0.017 0.003 0.001
(0.007) (0.020) (0.015) (0.007)

Number of observations 6934 3446 3633 6830


Log-likelihood 1851.160 1335.751 1116.921 1887.581
Note: Other controls include all the other variables presented in Table 4.

478 r 2011 European Association of Development Research and Training Institutes 0957-8811
European Journal of Development Research Vol. 23, 3, 459–487
Labour Market Outcomes

13 per cent if she lives in urban Bangladesh, compared with only 6 per cent if she lives
in rural Bangladesh. Part of this difference undoubtedly comes down to the availability
of employment opportunities for women in the rural areas. As already mentioned,
although there has been an expansion of female employment opportunities in Bangladesh,
a large part of this has arisen out of the expansion of the export-led RMG industry in
the urban centres of Dhaka and Chittagong. Despite this, employment is clearly available
for workers in the rural areas: an ‘average’ man has a probability of participation of
83 per cent in rural Bangladesh, compared with 93 per cent in rural Bangladesh.
We turn now to examine the effects of household poverty in this setting and once again
find evidence of a positive association between extreme poverty and participation in both
rural and urban areas, although the marginal effects are markedly different. Although
economic necessity is clearly greater in the rural areas (17 per cent of our rural sample is
from a household in extreme poverty, compared with just 8 per cent for our urban sam-
ple), being in a household in extreme poverty in urban Bangladesh raises a woman’s
probability of participation by 7 per cent, compared with only 3 per cent for a woman
living in the often poorer rural areas.
Reassuringly, we find that single women are still more likely than their married
counterparts to participate in the labour market, regardless of where they live. However,
living in urban Bangladesh raises a single woman’s probability of participation by
12 per cent, compared with just 5 per cent if she lives in a rural area. The traditional role of
a man as the ‘breadwinner’ of the family is also more pronounced in the rural areas than in
the cities.
Finally, although the presence of children has a more negative effect on the probability
of female participation in urban areas, the effects of having older people in the family is
more detrimental to female participation decisions in rural Bangladesh. This clearly
reflects attitudes towards families and their support networks. Women living in the rural
areas clearly have a better support network at their disposal. However, compared with
those living in the cities they are also expected to devote more time to the care of older
dependants. In addition, religion also has a role to play in rural Bangladesh, lowering a
woman’s probability of participation by 4 per cent, but is insignificant in the urban
centres.
There is thus little evidence, at least for the moment, that the change in attitudes
towards women’s employment in the cities has made its way to the rural areas.

Conclusion

Despite two decades of increased female labour market participation, we are yet to
fully understand how poverty (and perhaps more importantly extreme poverty) and
gender-related issues affect a woman’s participation in the labour market. This article
helps complement the prior often non-econometric-based evidence, by using data from a
large, nationally representative household data set for Bangladesh, a country that has
experienced rapid growth in female employment and changes in poverty, to model
employment outcomes for a representative sample of working-age adults. We address
several problems with existing studies in this area: notably small and selective samples,
and add to the prior literature in a number ways, but perhaps most notably by being able
to analyse labour market effects at the national level, by gender and across different
poverty states.

r 2011 European Association of Development Research and Training Institutes 0957-8811 479
European Journal of Development Research Vol. 23, 3, 459–487
Bridges et al

In doing so, we find that poverty is indeed important in explaining labour market
participation, and in line with other studies we find evidence of a positive association
between extreme poverty and participation. Such a finding persists after controlling for
personal/household characteristics and human capital variables and provides further
support for the view that a household’s economic choices are at least partially influenced
by the economic needs of the household and are not exclusively fixed by rigid societal
norms, although it appears that only certain types of occupations are open to the poorer
households.
However, these poverty-based explanations, although important, clearly do not explain
the whole story behind the growth in female employment in Bangladesh. Although Ban-
gladesh is still a deeply patriarchal society, there is growing evidence of a positive shift in
attitudes towards female employment by both households and employers. In particular,
there seems to be evidence of a growing acceptance of the participation of young single
women in the labour market. Such a finding persists after interacting marital status with
extreme poverty, which implies that it is not completely down to economic necessity,
although extreme poverty may have provided the catalyst. In addition, we find that male
self-employment is positively associated with the presence of larger numbers of children in
the household, which is perhaps indicative of slight changes to caring roles within families.
However, it is clear from the findings of this study that, although attitudes towards
women and work maybe slowly changing, there is still a long way to go. Rigid gender
norms still dictate patterns of employment among married men and women. This is
particularly true in the rural regions where traditional gender roles, such as the woman as
the primary caregiver, still hold.
Recommending policy measures while considering the tangled interplay of household,
work and gender dynamics is clearly not an easy task. Despite this, some concrete policy
recommendations follow from this and prior work. It is clear that action needs to be taken
to not only help improve the employability of women, but also to drastically improve the
wages and working conditions for those who do manage to find work. This is clearly a
complex, as well as a sensitive task, and one that cannot be dealt with through policy
interventions alone. Wage inequalities, in particular, not only arise because of gender
discrimination, but also arise because of differences in human capital endowments.
Moreover, private employers cannot be directly persuaded to increase the wages of their
female workers in the absence of enforceable legislation.
However, even considering these sensitivities, measures can still be taken. Most im-
portantly, perhaps, by enhancing the employability of women through greater access to
education, training and skills development, and by improving their access to inputs to self-
employment (that is, loans and grants). Indeed, Hulme and Lawson (2010) suggest that a
lack of appropriate working opportunities, or more specifically there being a lack of poor
work opportunities is one of the main traps that underpin extreme poverty. In the case of
Bangladesh, female education needs to be more closely linked to the needs of the job
market perhaps through awareness campaigns that highlight current skill shortages; young
women still have a tendency to restrict themselves to courses in ‘dressmaking and food
processing’, which clearly limits their employability (Rahman, 2005).
Better access to jobs and wages is also likely to improve the overall ‘bargaining position’
of women not only in the job market, but at home as well. In doing so, it will challenge the
widely held notion that women are simply secondary earners who can be paid less than
men, or that they can only do certain jobs and tasks, or work in certain industries, in
addition to helping confront other extreme poverty traps, such as social discrimination.

480 r 2011 European Association of Development Research and Training Institutes 0957-8811
European Journal of Development Research Vol. 23, 3, 459–487
Labour Market Outcomes

Acknowledgement
This research was undertaken with support from the ESRC’s Global Poverty Research Group
(GPRG) at the Universities of Manchester and Oxford (grant number M571255001). We are grateful
to Shanker Chandra Saha for assistance with data extraction. The authors are grateful for helpful
comments received from conference participants at EALE, 2009 and seminar participants at the
University of Nottingham.

Notes
1. We acknowledge as a side issue that long-term economic growth evidence indicates that women
are a smaller percentage of the labour force at moderate levels of development than at either
low or high levels of development – as recognized by prior papers such as Pampel and Tanaka
(1986) and Goldin (1994).
2. Only 42 women participate in the labour force for every 100 men (ILO, 2008).
3. The latest poverty headcount figures are from Serajuddin et al (2010).
4. Here, poverty is measured as per capita consumption below the poverty line, and extreme
poverty is per capita consumption below the lower poverty line. See, Serajuddin et al (2010) and
World Bank (2002) for further details.
5. The ILO defines the vulnerable employed as someone who is self-employed with no employees
(an own-account worker) or an unpaid family worker. The ILO argues that employment of this
kind is least likely than other forms of employment to have formal work arrangements, access
to benefits or social protection programmes.
6. In this instance, we make use of data from the nationally representative Household Income and
Expenditure Survey for 2000, which is a particularly rich data set for examining some of the
issues surrounding labour participation and employment outcomes. The quality of the
employment data is particularly good, relatively to other surveys, and the data set also provides
a detailed measure of the household’s poverty status, which is often not available in other
surveys.
7. A woman is thought to have her highest level of economic independence when she is engaged in
wage and salaried work or is an employer; lower independence when she is an own-account
worker (self-employed), and the least independence when she is an unpaid family workers (ILO,
2008).
8. Kantor (2003) and Anderson and Baland (2002) make similar observations for India and
Kenya, respectively.
9. See, for example, Mahmud (1997) for metropolitan Dhaka; Salway et al (2003) for the slum
dwellers in Dhaka; Kibria (1998) who interviews 70 garment workers and members of their
households; Kabeer and Mahmud (2004a) who conduct a comparison of women working for
the garment export sector and women working in the domestic market.
10. See, for example, Khandker (1987) for rural Bangladesh, Amin (1997) for two rural villages in
northern Bangladesh; and Anderson and Eswaran (2009) for married men and women in
Matlab, a predominantly rural area 70 km southeast of Dhaka.
11. Although ‘purdah’ literally means ‘curtain’ or ‘veil’, it is often used figuratively to describe this
custom of female seclusion.
12. Female-headed households are among the poorest in Bangladesh. Salway et al. (2003, p. 889),
for example, find that ‘ y many of the poorest households represent a breakdown of the
“patriarchal contract” y’ that is, are households without a male head.
13. A small proportion of garment factories in Bangladesh operate within the county’s export-
processing zones (EPZs). Kabeer and Mahmud (2004a) argue that, although there is a ban on
trade unions within the zones, wages and working conditions are generally better than
elsewhere.
14. Although there are more recent, nationally representative data for Bangladesh, there are issues
regarding labour supply data accuracy that reflect the usage of the HIES 2000 data for this
analysis, and given the poverty reduction that occurred through the 1990s the key examination
of poverty is further justified.

r 2011 European Association of Development Research and Training Institutes 0957-8811 481
European Journal of Development Research Vol. 23, 3, 459–487
Bridges et al

15. This involved calculating an upper and lower poverty line using food consumption and non-
food consumption data from the HIES (2000) – see, World Bank (2002) for further details.
16. Such a finding is in line with a report by the World Bank (World Bank, 2002), which uses the
2000 HIES data to show how participation rates vary across per capita expenditure deciles.
They show that participation is greatest among the poorest and richest deciles, although the
women tend to be employed in very different types of activities. They find that the majority of
women from the poorest households work for wages in agriculture, manufacturing or on family
enterprises, whereas women in the richest decile hold white-collar jobs, usually as teachers or
health-care workers.
17. See Paci and Sasin (2008) for a more detailed overview of the Bangladeshi labour market.
18. We test this assumption in what follows using a Hausman test and find insufficient evidence to
reject the null hypothesis that the errors are independent across alternative employment sectors.
It should be noted that our results are also robust to estimating a multinomial probit model,
which relaxes the independence of irrelevant alternatives assumption of the multinomial logit
model.
19. It should be noted that for completeness, the marginal effects are reported in the Appendix.
20. By ‘average’, we mean the probability of participation calculated using the mean characteristics
of the sample.
21. Although these results provide strong evidence for a direct link between poverty and
employment, there is, however, a potential bias in these estimates because of the likelihood of
unobserved characteristics that explain both poverty and participation. We address this
potential endogeneity by estimating the relationship between poverty and participation
simultaneously as a recursive bivariate probit, and find that our results are robust to this
potential bias (see, for example, Greene, 1997 and Bridges and Disney, 2010). These results are
not presented here but are available on request.
22. Indeed, interacting marital status with age we find that, not surprisingly, it is the young women
who are also single that are more likely to participate in the labour market (see Table A2).
23. The multinomial results are not reported here, but are available from the authors upon request,
and support the findings of the logistic regressions.

References
Afsa, R. (2003) Gender, labour market, and demographic change: A case study of women’s entry
into the formal manufacturing sector in Bangladesh. In: B. Garcı́a, R. Anker and A. Pinnelli
(eds.) Women in the Labour Market in Changing Economies: Demographic Issues. Oxford, UK:
Oxford University Press.
Amin, S. (1997) The poverty-‘purdah’ trap in rural Bangladesh: Implications for women’s roles in
the family. Development and Change 28(2): 213–233.
Anderson, S. and Baland, J. (2002) The economics of ROSCAs and intrahousehold resource
allocation. Quarterly Journal of Economics 117: 963–995.
Anderson, S. and Eswaran, M. (2009) What determines female autonomy? Evidence from
Bangladesh. Journal of Development Economics 90: 179–191.
Bangladesh CPR. (2005) Tales of Ascent, Descent, Marginality and Persistence. Chronic Poverty
Research Centre and BIDS, Manchester: University of Manchester.
Begum, S. and Sen, B. (2005) Maternal Health, Child Well-Being and Intergenerationally
Transmitted Chronic Poverty: Does Women’s Agency Matter? BIDS Working Paper, Dhaka,
Bangladesh.
Blackden, M., Canagarajah, S., Klasen, S. and Lawson, D. (2006) Gender and growth in Africa:
Evidence and issues. In: G. Mavrotas and A. Shorrocks (eds.) Advancing Development: Core
Themes in Global Development. Helsinki, Finland: WIDER.
Blunch, N. and Das, M. (2007) Changing Norms about Gender Inequality in Education: Evidence
from Bangladesh. Policy Research Working Paper Volume: 4404, The World Bank, South Asia
Region, Sustainable Development Department.
Bridges, S. and Lawson, D. (2009) A gender based investigation into the determinants of labour
market outcomes: Evidence from Uganda. Journal of African Economies 18(3): 461–495.

482 r 2011 European Association of Development Research and Training Institutes 0957-8811
European Journal of Development Research Vol. 23, 3, 459–487
Labour Market Outcomes

Bridges, S. and Disney, R. (2010) Debt and depression. Journal of Health Economics 29(May):
388–403.
Cagatay, N. and Ozler, S. (1995) Feminization of the labor force: The effects of long-term devel-
opment and structural adjustment. World Development 23(11): 1883–1894.
Glick, P. and Sahn, D. (1997) Gender and education impacts on the employment and earnings in
West Africa: Evidence from Guinea. Economic Development and Cultural Change 45: 793–824.
Goldin, C. (1994) The U-shaped Labor Force Function in Economic Development and Economic
History. NBER Working Paper Series, No. 4707. Cambridge: NBER.
Greene, W. (1997) Econometric Analysis, 3rd edn. London: Prentice Hall.
Hossain, N. (2004) Access to Education for the Poor and Girls in Bangladesh. World Bank.
Hulme, D. and Lawson, D. (2010) Making poverty reduction work for the poorest. In: D. Lawson,
D. Hulme, I. Matin and K. Moore (eds.) What Works for the Poorest: Poverty Reduction
Programmes for the World’s Extreme Poor. UK: Practical Action.
ILO. (2004) Global Employment Trends for Women 2004. Geneva, Switzerland: International Labor
Organization.
ILO. (2008) Global Employment Trends for Women 2008. Geneva, Switzerland: International Labor
Organization.
Kabeer, N. (1997) Women, wages and intra-household power relations in urban Bangladesh.
Development and Change 28: 261–302.
Kabeer, N. (2000) The Power to Choose: Bangladesh Women and Labour Market Decisions in London
and Dhaka. London and New York: Verso.
Kabeer, N. and Mahmud, S. (2004a) Globalization, gender and poverty: Bangladeshi women
workers in export and local markets. Journal of International Development 16: 93–109.
Kabeer, N. and Mahmud, S. (2004b) Rags, riches and women workers: Export oriented garment
manufacturing in Bangladesh. In: M. Carr (ed.) Chains of Fortune: Linking Women Producers
and Workers with Global Markets. London: Commonwealth Secretariat.
Kantor, P. (2003) Women’s empowerment through home-based work: Evidence from India.
Development and Change 34(3): 425–445.
Khandker, S. (1987) Labor market participation of married women in Bangladesh. The Review of
Economics and Statistics 69(3): 536–541.
Kibria, N. (1998) Becoming a Garments Worker: The Mobilization of Women into the Garments
Factories of Bangladesh. Occasional Paper 9, United Nations Research Institute for Social
Development, United Nations Development Programme.
Kishor, S. (2000) Empowerment of Women in Egypt and Links to the Survival and Health of their
Infants. In: H. Press and G. Sen (eds.) Women’s Empowerment and Demographic Processes:
Moving Beyond Cairo. Oxford, UK: Oxford University Press.
Lokshin, M., Glinskaya, E. and Garcia, M. (2004) The effect of early childhood development
programming on women’s labour force participation and older children’s schooling in Kenya.
Journal of African Economies 13: 240–276.
Mahmud, S. (1992) Women’s work in the family economy. In: H. Hossain, P.D. Cole and F.H. Abed
(eds.) From Crisis to Development: Coping with Disasters in Bangladesh. Dhaka, Bangladesh:
University Press.
Mahmud, S. (1997) Women’s work in urban Bangladesh: Is there an economic rationale? Devel-
opment and Change 28: 235–260.
Paci, P. and Sasin, M. (2008) Making Work Pay in Bangladesh: Employment, Growth, and Poverty
Reduction (Directions in Development). Washington, DC: World Bank Publications.
Pampel, F. and Tanaka, K. (1986) Economic development and female labour force participation:
A reconsideration. Social Forces 64(3): 599–619.
Paul-Majumder, P. and Begum, A. (2000) The Gender Imbalances in the Export Oriented Garment
Industry in Bangladesh. Policy Research Report, Working paper series 12. Washington DC:
World Bank Development Research Group/Poverty Reduction and Economic Management
Network.
Rahman, R. (2005) The Dynamics of the Labour Market and Employment in Bangladesh: A Focus
on Gender Dimensions. Employment Strategy Paper 2005/13. Geneva: International Labor
Organization.
Salway, S., Rahman, S. and Jesmin, S. (2003) A profile of women’s work participation among the
urban poor of Dhaka. World Development 31(5): 881–901.

r 2011 European Association of Development Research and Training Institutes 0957-8811 483
European Journal of Development Research Vol. 23, 3, 459–487
Bridges et al

Serajuddin, U., Zaman, H. and Narayan, A. (2010) Pro-poorest growth: A national household
survey approach. In: D. Lawson, D. Hulme, I. Matin and K. Moore (eds.) What Works for the
Poorest: Poverty Reduction Programmes for the World’s Extreme Poor. UK: Practical Action.
Standing, G. (1999) Global feminization through flexible labor: A theme revisited. World Devel-
opment 27(3): 583–602.
World Bank. (2002) Poverty in Bangladesh: Building on Progress. Poverty Reduction and Economic
Management Sector Unit, South Asia Region, (Report No. 24299-BD), World Bank,
Washington DC.
Yoshida, N., Narayan, A. and Zaman, H. (2007) Trends and patterns of poverty in bangladesh in
recent years. World Bank (mimeo).

484 r 2011 European Association of Development Research and Training Institutes 0957-8811
European Journal of Development Research Vol. 23, 3, 459–487
Labour Market Outcomes

Appendix

Table A1: Labour participation rate (%) by sex: Bangladesh

1983–1984 1984–1985 1990–1991 1995–1996 1999–2000 2002–2003 2005–2006


Women 7.9 8.3 14.1 15.8 23.9 26.1 29.2
Men 88.7 88.6 79.6 87.0 84.0 87.4 86.8
All 49.1 49.3 46.8 52.0 54.9 57.3 58.5
Note: Figures based on the Bangladesh Labour Force Survey (various years).

Table A2: Maximum likelihood estimates of the probability of participation: Interaction effects
(Marginal effects)

Variable Women Men Women Men


Household poverty status (base case top non-poor):
=1 if household extreme poor 0.034*** 0.054*** 0.054*** 0.021**
(0.01) (0.013) (0.013) (0.009)
=1 if household moderately poor 0.011 0.022*** 0.022*** 0.021***
(0.008) (0.007) (0.007) (0.006)
=1 if household mid-non-poor 0.011 0 0 0
(0.007) (0.005) (0.005) (0.005)

Personal characteristics:
Age 0.015*** 0.029*** 0.029*** 0.029***
(0.002) (0.001) (0.001) (0.002)
Age-squared 0.0002*** 0.0004*** 0.0002*** 0.0004***
(0.00002) (0.00002) (0.00003) (.00002)
=1 if single, 0 otherwise 0.083*** 0.110*** 0.110*** 0.080**
(0.012) (0.009) (0.009) (0.033)
=1 if divorced/separated, 0 0.146*** 0.088* 0.088* 0.054
otherwise (0.014) (0.047) (0.047) (0.179)
=1 if widowed, 0 otherwise 0.123*** 0.109*** 0.109*** 0.058
(0.01) (0.028) (0.028) (0.151)

Interactions:
Extreme poor  single 0.002 0.106*** — —
(0.024) (0.015) — —
Extreme poor  divorced/separateda 0.007 0.045 — —
(0.03) (0.06) — —
Extreme poor  widowed 0.025 — — —
(0.018) — — —
Age  single — — 0.003*** 0.001
— — (0.001) (0.001)
Age  divorced/separated — — 0.002 0.001
— — (0.001) (0.006)
Age  widowed — — 0.004*** 0.001
— — (0.001) (0.003)
Log-likelihood 3270.292 3005.350 3260.746 3028.168
Number of observations 10 380 10 463 10 380 10 463
a
For men divorced/separated also includes those that are widowed, because of small sample sizes.
Note: Other controls include all the other variables presented in Table 4.

r 2011 European Association of Development Research and Training Institutes 0957-8811 485
European Journal of Development Research Vol. 23, 3, 459–487
486
Table A3: Maximum likelihood estimates of labour market outcomes – Multinomial logit marginal effects

Variables Women Men


Bridges et al

Household poverty status (base case top non-poor):


=1 if household extreme 0.048*** 0.007 0.010*** 0.043*** 0.009 0.040*** 0.195*** 0.133*** 0.435***0.067***
poor (0.013) (0.004) (0.002) (0.011) (0.007) (0.008) (0.012) (0.008) (0.024) (0.018)
=1 if household 0.017** 0.003 0.007*** 0.018*** 0.004 0.036*** 0.158*** 0.072*** 0.298***0.032**
moderately poor (0.008) (0.003) (0.002) (0.005) (0.005) (0.007) (0.013) (0.01) (0.021) (0.016)
=1 if household 0.003 0.00001 0.005*** 0.006* 0.004 0.009 0.120*** 0.036*** 0.153*** 0.011
mid-non-poor (0.006) (0.002) (0.002) (0.003) (0.004) (0.006) (0.012) (0.01) (0.018) (0.013)

Personal characteristics:
Age 0.011*** 0.001*** 0.002*** 0.003*** 0.005*** 0.037*** 0.019*** 0.007*** 0.002 0.008***
(0.001) 0.00043 (0.001) 0.00049 (0.001) (0.002) (0.003) (0.003) (0.003) (0.003)
Age-squared 0.0002***0.00001***0.00002***0.0004***0.0001*** 0.0005***0.0003***0.00003 0.0001* 0.0001***
0.0002 0.00001 0.00001 0.00001 0.00001 0.00002 0.00004 0.00003 0.00004 0.00004
=1 if single, 0 otherwise 0.070*** 0.005 0.0002 0.026*** 0.039*** 0.168*** 0.084*** 0.012 0.065***0.030*
(0.015) (0.006) (0.006) (0.009) (0.01) (0.017) (0.018) (0.016) (0.019) (0.017)
=1 if divorced/separated, 0.226*** 0.014 0.01 0.073*** 0.130*** 0.133 0.163*** 0.177 0.059 0.089
0 otherwise (0.033) (0.01) (0.009) (0.017) (0.026) (0.132) (0.061) (0.113) (0.069) (0.071)
=1 if widowed, 0.158*** 0.022*** 0.020*** 0.062*** 0.055*** 0.254*** 0.022 0.004 0.169***0.103*
0 otherwise (0.02) (0.008) (0.007) (0.012) (0.014) (0.092) (0.075) (0.048) (0.034) (0.055)
=1 if sick, 0 otherwise 0.007 0.003* 0.002 0.001 0.002 0.009 0.011 0.012 0.014 0.006
(0.005) (0.002) (0.002) (0.002) (0.004) (0.007) (0.013) (0.011) (0.013) (0.013)

European Journal of Development Research Vol. 23, 3, 459–487


Household characteristics:
Age of head of household 0.001*** 0.0002*** 0.0001 0.0003*** 0.0001 0.0001 0.001** 0.001** 0.0001 0.0002
0.0002 0.0001 0.0001 0.0001 0.0001 (0.0003) (0.001) (0.0004) (0.001) (0.001)
Number of working-age 0.018*** 0.003*** 0.005*** 0.006*** 0.004*** 0.014*** 0.019*** 0.010***0.044*** 0.001

r 2011 European Association of Development Research and Training Institutes 0957-8811


adults in the household (0.002) (0.001) (0.001) (0.001) (0.001) (0.002) (0.003) (0.003) (0.004) (0.003)
Number of 0–5-year olds 0.012*** 0.0001 0.002** 0.004*** 0.010*** 0.002 0.031*** 0.024***0.041***0.015**
in the household (0.003) (0.001) (0.001) (0.001) (0.002) (0.004) (0.006) (0.005) (0.006) (0.007)
Number of 6–14-year olds 0.005*** 0.0004 0.001 0.002*** 0.003** 0.0001 0.018*** 0.014***0.025***0.008*
in the household (0.002) (0.001) (0.001) (0.001) (0.001) (0.002) (0.004) (0.003) (0.004) (0.004)
Number of adults above 0.020*** 0.005*** 0.005*** 0.005*** 0.004 0.010** 0.001 0.040***0.040***0.009
60 years of age in the (0.004 (0.002) (0.002) (0.002) (0.003) (0.005) (0.01) (0.008) (0.011) (0.01)
household
=1 if Muslim, 0 otherwise 0.034*** 0.009** 0.004 0.017*** 0.012** 0.005 0.092*** 0.049*** 0.016 0.022
(0.009) (0.004) (0.002) (0.004) (0.006) (0.008) (0.018) (0.011) (0.016) (0.015)

Highest level of educational achievement (base case no education):


=1 if primary incomplete, 0.003 0.001 0.002 0.003 0.003 0.016 0.010 0.021 0.043** 0.028
0 otherwise (0.009) (0.004) (0.004) (0.003) (0.007) (0.013) (0.026) (0.02) (0.019) (0.029)
=1 if primary complete, 0.016*** 0.001 0.004 0.011*** 0.007* 0.026*** 0.0003 0.015 0.090*** 0.101***
0 otherwise (0.006) (0.002) (0.003) (0.002) (0.004) (0.008) (0.018) (0.014) (0.012) (0.021)
=1 if secondary 0.027*** 0.001 0.002 0.009*** 0.015*** 0.086*** 0.023 0.013 0.200*** 0.124***
incomplete, 0 otherwise (0.005) (0.002) (0.003) (0.002) (0.004) (0.011) (0.014) (0.011) (0.01) (0.017)
=1 if secondary complete, 0.019*** 0.005** 0.004 0.008*** 0.002 0.166*** 0.065*** 0.058***0.256*** 0.214***
0 otherwise (0.007) (0.002) (0.003) (0.002) (0.006) (0.02) (0.017) (0.012) (0.009) (0.024)
=1 if high secondary, 0.011 0.002 0.007** 0.009*** 0.029*** 0.193*** 0.116*** 0.123***0.290*** 0.335***
0 otherwise (0.011) (0.003) (0.003) (0.003) (0.01) (0.02) (0.013) (0.009) (0.009) (0.023)
=1 if rural Dhaka, 0.019*** 0.005*** 0.016*** 0.003 0.027*** 0.009 0.017 0.202***0.091***0.085***
0 otherwise (0.006) (0.001) (0.005) (0.002) (0.003) (0.008) (0.016) (0.019) (0.012) (0.013)
=1 if urban Dhaka, 0.017** 0.001 0.004 0.004 0.024*** 0.058*** 0.094*** 0.110***0.133*** 0.091***
0 otherwise (0.008) (0.002) (0.003) (0.003) (0.006) (0.012) (0.02) (0.012) (0.014) (0.017)
=1 if rural Khulna, 0.032*** 0.003* 0.001 0.003 0.025*** 0.014 0.048** 0.285***0.095***0.156***
0 otherwise (0.006) (0.002) (0.004) (0.003) (0.003) (0.013) (0.02) (0.026) (0.016) (0.013)
=1 if urban Khulna, 0.014 0.007 0.009*** 0.018** 0.002 0.039** 0.200*** 0.108***0.083***0.048**
0 otherwise (0.011) (0.005) (0.003) (0.008) (0.006) (0.018) (0.03) (0.017) (0.02) (0.02)
=1 if rural Rajshahi, 0.0003 0.005*** 0.013*** 0.018*** 0.025*** 0.01 0.086*** 0.308***0.055***0.156***
0 otherwise (0.008) (0.001) (0.005) (0.005) (0.003) (0.009) (0.016) (0.021) (0.014) (0.012)
=1 if urban Rajshahi, 0.046*** 0.003 0.01 0.031*** 0.002 0.013 0.198*** 0.054***0.071***0.086***

European Journal of Development Research Vol. 23, 3, 459–487


0 otherwise (0.014) (0.004) (0.007) (0.009) (0.006) (0.013) (0.027) (0.019) (0.019) (0.017)
=1 if rural Barisal, 0.021*** 0.006*** 0.006 0.0001 0.021*** 0.026** 0.00003 0.157***0.082***0.102***
0 otherwise (0.007) (0.002) (0.005) (0.004) (0.004) (0.013) (0.022) (0.024) (0.017) (0.015)
=1 if urban Barisal, 0.011 0.002 0.004 0.006 0.002 0.039* 0.171*** 0.094***0.063** 0.053**

r 2011 European Association of Development Research and Training Institutes 0957-8811


0 otherwise (0.014) (0.003) (0.007) (0.008) (0.008) (0.021) (0.037) (0.021) (0.028) (0.024)
Number of observations 10 380 10 380 10 380 10 380 10 380 10 463 10 463 10 463 10,463 10 463
Labour Market Outcomes

487
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Potrebbero piacerti anche