PreprintPDF Available

Urban Female Labor Force Participation and its Correlates: A Comparative Study of Slum-dwellers and their Urban Counterparts of Three Metro Cities in India

Authors:
Preprints and early-stage research may not have been peer reviewed yet.

Abstract and Figures

Forthcoming in "Advances in Women’s Empowerment: Critical Insight from Asia, Africa and Latin America Vol: 29", Emerald UK. . This paper deals with an important but neglected aspect of female labor force participation (LFP) in urban India. Contemporary literature that deals with urban female LFP in India typically focuses on the entire urban sector, and invariably misses out one important aspect of urban living-the slums and its dwellers. This study attempts to fill that critical gap by examining two different household surveys side-by-side: a primary survey particularly focused on households living in slums and slum-rehabilitated colonies and the nationally representative Indian Human Development survey-II. It brings outs a comparative picture of nature, type jobs that women take up and various correlates of female LFP from both slum-and non-slum areas of three metro cities of India, viz. Delhi, Kolkata and Mumbai. It further explores the similarities and differences of the correlates for female LFP among the slum-clusters of these cities. It is found that despite being poorer and marginalized in many ways the slumdwelling women's work participation rate is not extraordinarily high compared to their non-slum urban counterparts. In slums, a higher percentage of women are engaged in self-employment (including family business) and casual employments (includes domestic helps), whereas in non-slum areas somewhat more women are engaged in regular salaried jobs. Analysis identifies correlates that have similar effects across slum and non-slum areas-relationship between education and FLFP reflects a flat-bottom J-shaped pattern; being married, higher child dependency ratio and household heads with higher education significantly constrain women's work choice; strong income effect of other household members earning on FLFP, but asset holding has no bearing. However, there are other factors that affect FLFP differently in slums and non-slum areas. These findings help us drawing important policy prescriptions.
Content may be subject to copyright.
Urban Female Labor Force Participation and its Correlates: A Comparative Study of
Slum-dwellers and their Urban Counterparts of Three Metro Cities in India
Sugata Bag
This version: January 2020
Forthcoming in 2020 in Emerald series: Advances in Gender Research
Issue: Advances in women’s empowerment: critical insight from Asia, Africa and Latin America
Abstract
This paper deals with an important but neglected aspect of female labor force
participation (LFP) in urban India. Contemporary literature that deals with urban female
LFP in India typically focuses on the entire urban sector, and invariably misses out one
important aspect of urban living – the slums and its dwellers. This study attempts to fill
that critical gap by examining two different household surveys side-by-side: a primary
survey particularly focused on households living in slums and slum-rehabilitated colonies
and the nationally representative Indian Human Development survey-II. It brings outs a
comparative picture of nature, type jobs that women take up and various correlates of
female LFP from both slum- and non-slum areas of three metro cities of India, viz. Delhi,
Kolkata and Mumbai. It further explores the similarities and differences of the correlates
for female LFP among the slum-clusters of these cities. It is found that despite being
poorer and marginalized in many ways the slumdwelling women’s work participation rate
is not extra-ordinarily high compared to their non-slum urban counterparts. In slums, a
higher percentage of women are engaged in self-employment (including family business)
and casual employments (includes domestic helps), whereas in non-slum areas somewhat
more women are engaged in regular salaried jobs. Analysis identifies correlates that have
similar effects across slum and non-slum areas – relationship between education and
FLFP reflects a flat-bottom J-shaped pattern; being married, higher child dependency
ratio and household heads with higher education significantly constrain women’s work
choice; strong income effect of other household members earning on FLFP, but asset
holding has no bearing. However, there are other factors that affect FLFP differently in
slums and non-slum areas. These findings help us drawing important policy prescriptions.
Acknowledgements: This paper took shape during my tenure as ICCR Visiting Chair at the Centre for
Contemporary Indian Studies, University of Colombo. I thank Indian Council for Cultural Relations (ICCR) for this
opportunity.
This work is part of the Nopoor Research Project supported by the European Union (www.nopoor.eu) Seventh
Framework Programme (FP7/2007-2013) under grant agreement n°290752.
Department of Economics, Delhi School of Economics, sugata@econdse.org.
1
1. Introduction
Women’s role in economic development has long been recognized squarely by the scholars and
the statesmen. According to Lewis (1954), the transfer of women’s work from household to
commercial employment is one of the most notable features of economic development.
Jawaharlal Nehru, former Prime Minister of India, once remarked, I have long been convinced
that a nation’s progress is intimately connected with the status of its women(Parthasarathi &
Nehru, 1985). Women’s participation in employment is an important indicator, although
imperfect, for women empowerment, as it can help reduce gender inequality. However, Indian
labor market has been historically characterized by low female labor force participation (FLFP).
The labor force participation rate (LFPR) of females in India lags considerably behind the
international “norm” 1 (Bhalla and Kaur 2010) and has even fallen over the past decades. It is the
urban component that is particularly very low and remained stagnant over a very long period
(Klasen and Pieters 2015).
Contemporary literature that deals with urban FLFP in India typically focuses on the entire urban
sector, and invariably misses out one important aspect of urban living – the slums and its
dwellers. One reason might be lack of data, incidentally most of the existing studies are based on
the nationally representative surveys that neither collect enough information specific to slums
nor even have slum identifiers.2 Thereby naturally, there exists no study on FLFP among the
slum-dwellers of urban areas, or any comparison vis-à-vis their non-slum counterparts, although
a sizeable proportion of the urban population is living in slums which are distinct from the rest of
the city dwellings in terms of worse living conditions, socio-economic environment. Table 1
below shows that over a third of India’s urban households (38%) lives in its fortysix “Million-
plus” cities. Of the six metro cities of India, three metros Mumbai, Delhi and Kolkata - have
significantly higher slum population than others, not only in absolute number but also in
1 Globally, FLFP has remained relatively stable in the two decades (1990-2010), at roughly 52% (KILM-ILO 2014).
2 For example, Employment and Unemployment Surveys by National Sample Survey Organisation (NSSO).
However, other national level survey, viz. National Family Health Surveys (NFHS), although contains slum
identifiers, it does not collect any information on monetary indicators. Moreover, both Agarwal (2011) and Carr-Hill
(2013) point out that the nationally representative household surveys are not appropriate for obtaining information
about slum populations and other disjoint populations.
2
proportion.3 Bag, Seth, and Gupta (2019) noted that the slums in these three metro cities existed
since the colonial days and the people had been living there over many generations.
Table 1: Distribution of Slum Population in Million Plus Indian Cities
Million Plus
cities in India
Total Popn.
(in ’000)
Slum Popn.
(in ’000)
% of Slum to
Total Popn.
Proportion of Slum
HHs to Total Urban
HHs (
%)
*
11,978
6,475
54.06
41.3
Delhi
*
9,879
1,851
#
18.74
14.6
Kolkata
*
4,573
1,485
32.48
29.6
Chennai
*
4,344
820
18.88
2
8.5
Hyderabad
*
3,520
627
17.23
31.9
Surat
2,433
508
20.89
-
Pune
2,538
492
19.39
-
Ahmedabad
3,637
474
13.46
-
Bangalore
*
4,301
431
10.02
8.5
Kanpur
2,551
368
14.42
-
All India
73,346
17,697
24.13
38%
Source: This table is reproduced from Bag, Seth, and Gupta (2019), compiled from Census of India, 2011.
Notes: * Metro cities.
#
In Delhi, apart from the slums, there exists two types of slum rehabilitation settlements, called
“Resettlement” (45) and “Relocation” (37) colonies. Together they host about 2 million dwellers
(authors’
calculation, based population growth since 2001). These two types of habitations however are not
legally
considered as slums, but living conditions are much similar to the slums in Delhi (see,
Bag, Seth, and
Gupta
(
2019)
)
.
The aggregate urban trends noted in the existing studies mask large differences in nature and
rates of female workforce participation across cities, across metros and non-metros, within urban
sector slum and non-slum areas. Therefore, there is a gap in our knowledge base about a
significant but vulnerable and often marginalized section of urban dwellers that live in slum.
From the policy perspective as well, studies are required at a more disaggregated level, to
capture the diversity in the rate and the type of work participation of women and their
determinants across slum and non-slum areas within urban areas.
This paper is a small but significant attempt to fill this crucial gap in literature. We focus mainly
on three metro cities with highest slum population in India. Using a primary household survey
data on the three metro cities Delhi, Kolkata and Mumbai, this paper first investigates the rate
3 Worth noting that the Primary Census 2011 Abstracts states that the total slum population in India has reached 65.5
million (about 22% of the total urban population) in 2011, with an astounding decadal growth rate of 25%. Roughly
13.7 million, or 17.4%, urban Indian households lived in a slum in 2011, i.e. nearly one in every six Indian urban
residents lives in a slum.
3
and the nature of women’s (aged between 18-60) participation in the labor market. We then bring
out a comparative picture of work participation pattern and rate between slum and non-slum
areas of these cities using other secondary databases from comparable time frame. Issues
surrounding the methodology and underestimation of women’s work participation are also
tackled. We show that although participation rate (using various definitions) in slum areas are
higher than the non-slum areas, but the difference in LFPR is not so huge. However, there is
wide variation in the type of jobs women undertake in slums and non-slum areas. Next, using a
simple labor supply model, we estimate the determinants of LFP of women residing in slums and
their urban counterparts at the three-city aggregated level, separately. We take a step further and
look at comparative analysis the determinants of FLFP in slum-clusters of the individual cities.
The key contribution of this paper is that our estimation results provide a detailed account of a
comparative vision on how the work-choice of females and its correlates, in urban slum of three
the three metro cities differ from their non-slum urban counterparts. Important similarities are
identified across slum and non-slum areas – relationship between education and FLFP reflects a
flat-bottom J-shaped pattern; being married has very similar strong negative effect, child
dependency (below 6 yrs.) significantly constraints work choice, strong income effect of other
household members earning on FLFP but asset holding has no bearing.
The paper is structured as follows. Next section briefly reviews the findings of existing literature.
Section 3 describes the datasets that are used in this analysis. In section 4, we provide some
stylized facts on workforce participation rates and pattern of occupation. Section 5 describes the
labor supply model, construction of variables and also provides some descriptive statistics.
Section 6 brings out a comparative analysis of the determinants of female labor force
participation of aggregated slums and non-slum areas of three cities. In a subsection, we delve
deeper into the city specific slum level determinants, separately for three cities. Section 7 draws
conclusions for this analysis. An appendix is provided at the end, that contains various
comparative figures related to FLFP.
2. Brief Literature Review
There exists an extensive literature on the FLFP and its determinants in Indian context some
focus at the pan-India level, others at the sub-national levels of rural and urban. These studies
4
highlight several supply-side, institutional and demand-side factors. Notably there exists no
study on the urban slumdwellers.
We can divide existing literature into two main strands. The first strand considers how
demographic characteristics and educational attainment affect the LFP decisions of women. The
second strand of literature, the less trodden one, deals with demand-side factors such as wage
and rigidities in labor markets. The studies under second strand noted a shortage of medium-
sized enterprises in India, and have linked firm hiring decisions to cross-state differences in labor
market regulations, which is also a reason for high share of informal employment in overall
employment (Das et al. 2015; Gonzales et al. 2015). Labor market demand-side factors— e.g.
gender gap in wage, lack of good jobs for women and process of urbanization—are causing low
female participation (Chatterjee et al. 2015; U. Chatterjee, Murgai, and Rama 2015; Bhalla and
Kaur 2010; Kingdon and Unni 2001; Agrawal 2014).
Studies from the first strand noted that contrary to the Neo-classical theory of monotonically
increasing FLFP along with women’s education attainment, there exists a unique U-shaped
relationship in India (Reddy 1979; Das and Desai 2003; Kingdon and Unni 2001; Das 2006;
Klasen and Pieters 2015, Andres et al. 2017; Kanjilal-Bhaduri and Pastore 2018; Chatterjee,
Desai, and Vanneman 2018), similar to many developing countries. This U-shaped relationship
is, on one side, distress-driven for poorly educated women who are obligated to work to support
themselves and families; on the other side, stigmas attached to taking up low-skilled or menial
employments for the moderately educated women; on the far extreme, there exists attractive job
opportunity with higher wages that induces better-educated women to work. Moreover, an
increase in income of other household members significantly reduces the probability of woman’s
employment (E. Chatterjee, Desai, and Vanneman 2018; Sarkar, Sahoo, and Klasen 2019).
Women’s, particularly married ones, decision to participate depends largely on the presence of
young children (below 5), and the number of elderly dependents in the household, disappearance
of multi-generational households over time coupled with lack of crèches and institutional child
support for working young mothers contribute to the low or decline in female LFPR (Das 2006;
Rani and Unni 2009; Bhalla and Kaur 2010; Sengupta and Das 2014; Kapsos et al. 2014; Das et
al. 2015; Sorsa et al. 2015; Das and Žumbytė 2017).
5
Women’s LFP is also determined to a large extent by caste, religion, social status, marital status,
and other socio-cultural norms (Eswaran, Ramaswami, and Wadhwa 2012; Klasen and Pieters
2015; Nicholas and Srinivas 1967). These factors operate at multiple levels in society and restrict
women’s mobility and access to wage employment, thereby forcing women to often take up non-
wage employment, or to remain out of the labor force (Das 2006; Sethuraman 1998; Ghosh
2009; Desai and Jain 1994; Thomas 2012). In Indian set up, women are held primarily
responsible for household duties, care-giving and reproduction; further, male members of the
household, particularly husbands, generally decide on what type of job women should take up
(Sudarshan and Bhattacharya 2008). It is also noted that social, cultural, historical, and economic
factors all play a role in determining the pattern of occupational segregation, therefore women
crowd into certain jobs which are low in the occupational hierarchy, payment and status, but are
considered socially acceptable (Swaminathan and Majumdar 2006; Rustagi 2013).
3. Choice of Datasets
This analysis draws upon two different household surveys: a primary survey (NoPoor) for the
slum dwellers, and a secondary database (Indian Human Development Survey-2, hereafter
IHDS-2) for the other urban (loosely called non-slum) dwellers of the three metro cities of India.
These two databases were compiled within a small gap of two years. We use these two databases
mainly because of comparability and richness of individual and household levels available
information.
The data for the slumdwellers comes from the primary household survey data that we collected
in 2013-14 through two-stage stratified sampling from the slums of the municipal corporation
areas of Kolkata, Mumbai and Delhi.4 The sample size from each city was based on the total
slum population size of the city as well as on the degree to which regional or other subsamples
representations were required. In Kolkata, from 15 boroughs we randomly selected 63 slums,
interviewed 808 households. In Delhi, from 11 revenue districts we randomly selected 57
squatter-settlements and interviewed 854 households. In Mumbai, from 23 wards we randomly
selected 96 slums and interviewed 1,228 households. Further, the authorities of Mumbai and
4 Survey conducted as part of the European Union funded global research project NOPOOR. For further details on
sampling design, see Bag, Seth, and Gupta (2016) and Bag, Seth, and Gupta (2019)
6
Delhi envisaged into slum relocation and redevelopment program for quite some time.5 For the
sake of capturing the diversity, we have collected samples from these localities as well – 27
resettlement/relocation colonies of Delhi were randomly selected, and interviewed 431
household, and in Mumbai the residents of the SRA buildings from surveyed slum locations
were also included. Including all these slumdwellers from three cities and resettled dwellers from
Delhi and Mumbai, our sample size stands at 5342 females in the age group 18-20 years from
3168 households, however we exclude households that do not have female members.
The IHDS-2 is a nationally representative database, conducted during 2011-12, spread across all
the States and Union Territories of India.6 The sample covers 42,152 households (205,002
individuals) from 384 districts, 1503 villages and 971 urban blocks. However, for our purpose
we use only a small subset, relevant for the three metro cities, Delhi Kolkata and Mumbai,
comprising about 1386 households and 2123 females in the age group 18-20 years. Note that
about 2% of urban household samples in three metro cities can be identified as slum dwelling
housing units, by dropping them we ensure that our sample only comprises of non-slum houses.
Both the databases are multi thematic surveys, which have collected individual and household
level information. NoPoor focuses more on slum, while IHDS takes care of pan-India diversity.
At the household level, information collected on religion, caste, type of housing and ownership,
assets, land, access to basic civic facilities and government schemes, incomes from assets,
consumption and expenditure details on basic food items. For individuals, we have information
on age, gender, marital status, age at marriage, literacy and educational details, migration details,
employment details including information of earning and past occupations, savings and
insurance details, health related information.
5 Delhi has a long history of slum relocation and rehabilitation since 1970s. “It was claimed that the attempt was to
make an all-round development for the relocated ‘squatters’ and that is why it was ‘resettlement’ and not just
‘rehousing’. But the present-day ground reality poses a different picture that can only invoke pity” (Dupont 2008).
In Mumbai, Slum Rehabilitation Authority (SRA) since early-2000s has been working on in-situ slum
redevelopment programme. Many multi-story buildings (known as SRA buildings) have been constructed in the
place of slums to rehabilitate original slum dwellers.
6 This dataset is produced by the National Council of Applied Economic Research (NCAER), New Delhi, and the
University of Maryland.
7
4. Workforce participation – definitional issue, some stylized facts
The labor force participation rate is a measure of the proportion of a country’s working age
population (i.e. aged 15 to 65) that engages actively in the labor market, either by working or by
looking for work (ILO KILM, 8th Edition).7 Contemporary literature points out that the
estimating labor participation is not an easy task, there exists a possible under-measurement of
women’s work especially in developing countries (Kapsos, Silberman, and Bourmpoula 2014;
Sudarshan and Bhattacharya 2009; IAMR and ILO 2013; Hirway and Jose 2011). The
convention in most developing countries is to measure labor force related variables on the basis
of weekly status, e.g. “did you work for at least one hour in the preceding week”, accordingly
occupation, industry and wage in the previous week are then recorded. However, in India, there
are three (and more) definitions of labor force.
Most frequently used NSSO (National Sample Survey Organisation) definition of employment is
a combination of Usual Principal Status (UPS) and Subsidiary Status (USS). The prevalence of
these two definitions for work status emanates from the structure of the economy in the 1950s
and 1960s when it was majorly agricultural. An individual is defined as being employed –
according to UPS, if the person engages in the economic activity for a “majority of the year”;
whereas according to USS, if person engages in an economic activity for “at least 30 days” in the
365 days preceding the survey.8 Often an individual engages into multiple activities in different
phases during the preceeding year, and depending upon the duration, activities maybe classified
in one or other usual status activity; therefore combining UPS and USS gives us slightly higher
LFP rates.
However, the IHDS-2 measure of workforce participation is more detailed than other surveys,
such as the Employment and unemployment surveys by NSSO (Chatterjee, Desai, and
Vanneman 2018). IHDS-2 has separate modules for three different types of work (e.g., on
household farm and non-farm businesses, and wage labor); it asks which household members
participated in each type of work, then number of days worked in the preceding year, and hours
worked in a day in each occupation are recorded. Using this, total hours (or total days) worked
7 Online at: http://www.ilo.org/empelm/what/WCMS_114240/lang--en/index.htm
8 Further, since the late 1970s, the NSSO started collecting data on the work status on a daily basis, and within each
day, as well as on a half-daily basis.
8
across all categories in the preceding year is computed.9 To be noted that the way our NoPoor
survey captures work related information, it is very comparable to IHDS, rather than NSS.
Moreover, unlike NSSO, both of the surveys do not properly account for persons who are
currently unemployed but looking for job.
The IHDS has two definitions of being employed, these versions interpret the economic activity
individuals for the “majority of the year” differently.10 An individual is considered employed in
the preceding year (definition 1, officially adopted by IHDS) if worked at least for 240 hours;
and (definition-2) if worked for at least 180 days. This definition-2 (i.e. 180 days per year) is
comparable to the ‘usual principal status’ definition used by NSS at the national and subnational
levels of rural and urban, but may not so at a much smaller level (say, urban metro cities).11
However, the definition-1 (240 hours per year) captures the fragmented and multiple activities in
greater way than 30 day cut off NSS USS, thereby shows work participation rates for women are
higher than the NSS participation rates, nationally and sub-national levels of rural and urban, but
again at lower level comparability remains a concern. Therefore, we also construct LFP rates
using a far more relaxed criterion of working any hour (or days) in the preceeding year.
We thereby construct the Table 2 below, using the aforementioned definitions of work without
accounting for ‘looking for job’ component, and call it workforce participation. We use IHDS-2
for the cities, and NoPoor databse for slums of three metro cities, Delhi, Kolkata and Mumbai.
We also use NSS Employment and Unemployment Survey 68th round (2011-12) for basic
reference. An interesting pattern is visible between the participation rates captured by NSS-68
and IHDS-2. For the three metro cities, the LFP rates for male, female or overall, aged between
18-60, captured by NSS are systematically higher than that of IHDS, however the gap in LFP
rates across the two surveys seems to narrow down for the superset of six metros. This could be
related to sampling issues and may require deeper investigation, but we move on as our focus is
on the slumdwellers.
9 To be noted that in many cases, individuals are, either periodically or simultaneously, employed in multiple works.
10 Following Psacharopoulos and Tzannatos (1989), our definitions of LFP excludes the ‘domestic labour’ i.e. work
done by people for themselves and for other household members. Indian women often remain engaged into caring
for household animals, collection of firewood or other fuels, and fetching water from public sources, but were not
included as LFP as these are usually regarded as normal household chores in India.
11 Definition-2 somewhat differs from the NSS definition of principal status but gives similar estimates at sectoral
levels. The use of 180 days as an approximation of at least 50% of days worked in a year which is similar to the
“major time criterion” used by NSS to define work status. Klasen and Pieters (2015) and Desai_et_al._ (2009)
supported this.
9
Table 2: Workforce Participation Rates for Persons (aged 18-60)
Survey
Work
Definition
Urban 6 Metros Urban 3 Metros
Overall Male Female Overall Male Female
NSS 68
UPS
54.19%
84.29%
21.27%
54.78%
83.61%
20.27%
NSS 68
UPS/USS
55.14%
84.73%
22.77%
56.17%
84.16%
22.53%
IHDS 2
180
d
ay/yr
45.11%
73.20%
15.93%
43.96%
71.48%
14.70
%
IHDS 2
240
h
r/yr
48.29%
76.31%
17.73%
45.95%
74.18%
15.
87
%
IHDS 2
A
ny
wk/yr
48.29%
76.87%
18.58%
46.45%
74.62%
16.49%
NoPoor
180
d
ay/yr
50.04%
75.36%
23.34%
NoPoor
240
h
r/ yr
48.63%
73.00%
22.95%
NoPoor
A
ny
wk /yr
5
2.36%
78.49%
25
.
39
%
Note 1: Author’s calculation from NSS 68th round Employment and Unemployment Survey (2011-12), IHDS 2
(2011-12) and NoPoor Survey (2013-14). Individuals looking for job are not included.
Note 2: Primary Census 2011 Abstracts notes that Workforce participation rates in slums are 36.4% for all
persons, 54.3% for males and 17.1% for females in the working age group 15-65.
Female LFPR in slums (NoPoor) of three metro cities hovers around 23% (according to def-1 or
def-2), which is very low relative to their male counterparts. Further, it is to be noted that the
female participation rates according to NSS (urban 2011-12) and NoPoor (slum urban 2013-14),
for various definitions in three metro cities, are very comparable, but for IHDS (non-slum 2011-
12) rates are tad lower. However, using our third definition (work any hr or days in preceeding
year), we see a small jump in LFP rate of women to 25.4% in slums. This points to the fact that
there exists only a handful of women that work less number of days or hours in a year. One
important observation is that despite being poorer and marginalized in many ways the slum
dwelling women’s work participation is not extra-ordinarily high compared to their rest of the
urban counterparts. Perhaps the slum dwellers in urban areas closely follows the others.
Let us now briefly look at the occupational patterns in slums and non-slum areas of these metros.
Table 3, using work any hr definition, below shows that the variation is quite stark. In slums
(NoPoor) a higher percentage of women, about 80% of the total employed, are engaged in self-
employment (including family business), and casual employments, whereas in non-slum areas
(IHDS-2) about 58% women are engaged in regular salaried jobs (long or short contracts in
government or private sectors). Disproportionately higher percentage of casual workers (includes
domestic helps) in slums reflects economic vulnerability and lower education levels among the
women there.
10
Table 3: Occupational Pattern for Women (aged 18-60) in Three Cities
across Slum and Non-slum Areas
Type of Work IHDS-2
(
Non
-
slum
)
NoPoor
(
S
l
u
m
)
Self Employed + Family Business/farm
4.05%
8.20%
Casual Labor
2.87%
12.13%
Regular Salaried… (9.56%) (5.08%)
…Government Permanent
2.73%
0.90%
…Government Contractual
-
0.79%
…Private Contract (tenured)
1.37%
1.03%
…Private Contract (non-tenured)
5.46%
2.36%
Observations
2,114
5,330
Workforce Participation rate (any work/yr) 16.49% 25.39%
5. The Model and the Covariates
To determine the covariates of workforce participation decision of women in urban slum and
non-slum areas, we use the standard supply-side analytical framework which emphasizes on the
characteristics of the woman herself and those of her household. Using aforementioned two
datasets, i.e. IHDS-2 for non-slum and NoPoor for slums in those metro cities, we aim to test
how similarly or differently the individual and household level factors, as highlighted in the
existing literature, have contributed towards the low level of FLFP in the slums and non-slum
areas of the three urban metro cities of India. We further delve into segregated city level analysis
for the slums. The framework takes the form of a series of nested specifications for the decision
to work by women between 18-60 years of age.
5.1. The Model
The probability of woman j from household h, and in city k being in the labor force (including
self-employment, unpaid family work, regular and casual wage employment) is modelled as:
𝑃𝑟
(
𝑊
)
=
𝐹
𝐼
,
𝐻
,
𝐶
,
𝜀
(1)
where: dependent variable, working status of the woman 𝑊
 , is dichotomous and assigns the
value one if a woman was employed last year, and zero if that is not the case (following either
definition 1 or 2 of working). I
jh
k is a vector of the characteristics of the woman making the
11
decision. 𝐻hk is a vector of the characteristics of her household, and of the head of household.
The city fixed effect is captured by 𝐶. Lastly, 𝜀 is an independently distributed stochastic
disturbance with zero mean.
5.2. Variable Construction, Sample Means and Descriptive
Most of the variables constructed are standard in the literature. We have several woman’s
individual level variables. Woman’s marital status is captured through a categorical variable – single,
currently married, and currently widow/separated/divorced. Dummies for poor health, and migration
status from other town or villages of woman are controlled. Instead of using woman’s years of
schooling directly, a 7-category variable is constructed (illiterate, literate (0 years schooling), primary
(1-4), middle (5-8), secondary (9-10), higher secondary (11-12), and graduate or above ≥ 12).
To capture how women negotiate with care responsibilities within household (towards children and
elderly), three variables are constructed. As it is not clear, the care-giving responsibility rests on
whom – daughter or daughter-in-law, we created two ratios of total number of children below 6
years, and also between 6-14 years, to total number of women in the age group 18-60 years. As these
ratios go up, it is expected to reduce women’s LFP. Similarly, we have a ratio of total number of
elderly people (≥ 65 years) to total women between 18-60 years within the household. Presence of
elderly people is expected to ease the burden of child rearing for other women in the family, but that
may increase the care burden directed to them.
To demonstrate the position of woman within household and vis-à-vis its head, we have several
variables that capture the household’s and its head’s characteristics. We have an indicator if the
household is headed by female.12 We are interested in the role of women’s own versus household
head’s education and age, among others. As we expect a non-linear effect, a categorical variable for
head’s age is controlled (5 categories: 30 yrs, 31-40, 41-50, 51-60, and 60 yrs). We have two
different expectations: either older heads to be more conservative, or the heads in their economic
prime (aged 40-60 yrs) may hold strong patriarchal views. A four-category variable for head’s
education level is used (illiterate, literate–primary, middle–higher secondary, graduate and above)
with the expectation that higher education brings forth more liberal values, however households with
low-educated heads reflects lower economic strength.
12 Often it is noted that female headed households are vulnerable as these women are most likely to be widowed,
divorced or separated, and their FLFP is higher.
12
The household size and its square are used to capture any non-linear effect of it. In Indian context,
household’s social status, reflected by the caste affiliation of the household, plays a major role in
FLFP (Eswaran, Ramaswami, and Wadhwa 2012). A six category religion-caste combined variable is
constructed that may capture direct fixed effect of culturally or religiously determined restrictions on
women, which are expected to be strongest among Muslim and high-caste Hindu households (Chen
and Dreze 1992; Das and Desai 2003). Categories are: Hindu Upper Caste, Hindu Other Backward
Classes (OBC), Muslim, a combined category for all Scheduled Caste (SC) and Scheduled Tribes
(ST) irrespective of their religious subscription, and all other minority religious groups (such as
Buddhist, Christian, Sikh, Jain, Parsi) clubbed together.13
And to reflect upon the so-called wealth effect and income effect, we construct two variables.
Although both asset holdings and income broadly reflect the standard of living of a household, they
may not measure the same aspect of it. To get around the endogeneity issue of using total household
income directly, we constructed ‘real per capita income earned by other household members’ i.e.
an individual’s own income is subtracted from total household income from all sources, then divided
it by the household size to get the ‘per capita’ estimate, and use available deflator to convert it into
‘real’.14 Further, as our variable has zeros and negative values (owing to business loss), we created 4
quartiles for each cohort of slum dwellers and their counterparts.
An asset index score is calculated, to reflect wealth level of the household, separately for each
data cohort, using principal components analysis, based on the commonly available factors in
two of our datasets – consumer durable goods (such as television, fridge, telephone, mobile,
motorcycle, car, washing machine), and housing factors (such as whether improved housing
materials used for roof, floor and wall), and also the access to utilities (such as separate kitchen,
personal toilet, piped water and electricity in the premise, and LPG gas cylinder). See, Table 8 in
appendix. As we expect the effect of wealth index to be non-linear, we divide the scores of
wealth index into 4 quartiles for relevant cohorts.
Finally, we also include the city dummies to take care of the other unobservable heterogeneous
effects across the cities.
13 The combined SC and ST category is created due to lack of numbers for ST households in the metro cities. An SC
or ST family could have their religious belief in Hinduism, Islam or any other faith, but all are clubbed together.
Literature often points out that both SCs and STs are at the bottom of the social hierarchy.
14 We follow the method proposed by Chatterjee, Desai, and Vanneman (2018) and Sarkar, Sahoo, and Klasen
(2019) to calculate ‘other family income’, but go a step further to calculate it at per capita level by dividing that by
household size, which is better estimate of economic status and total income excluding the woman’s own income.
13
5.3. Summary statistics
The two tables below, one for household level factors and other for women’s individual level
factors, show the unweighted sample means of the variables from two databases and their
differences. Let us highlight some important observations.
Table 4 shows that there is higher presence of SC/ST and female headed households in the slums
(especially in Delhi). Heads’ age distribution is roughly similar, however there exists significant
differences in distribution of heads’ education across slums and rest of the urban areas – a very
high percentage of household-heads in slums are illiterate, and a very small percentage attended
college. Household size is significantly higher in slums. Different child dependency ratios are
similar, but slums have higher old-age dependency. Slum dwellers have significantly lower asset
base, moreover their monthly per capita earnings are also much lower. Migration pattern also has
considerable differences among slums and non-slum areas when in-migration has fallen after
1990 compared to 1970s or 1980s in slums (decadal share/percentage of migration of total
sample), the in-migration to metro cities non-slum areas witnesses a surge in post 1990.
Table 4: Sample Means for Household level variables
in three Metro Cities (percent)
Variables IHDS2
(1)
NoPoor
(2)
Difference
(1) – (2)
Social Groups: Hindu Upper Caste
0.41
0
.29
0.12***
… Hindu OBC
0.19
0.14
0.05***
… Muslim OBC
0.04
0.05
-
0.01
… Muslim Non
-
OBC
0.08
0.15
-
0.07***
… Scheduled Caste/Tribes
0.25
0.35
-
0.09***
…. Other Minority Religions
0.02
0.02
0
Head: Female
0.13
0.18
-
0.05***
Head's Age (years)
$
49.28
48.57
0.71
Head's Age: below 31
0.05
0.06
-
0.01*
… 31
-
40 yrs
0.2
0
0.22
-
0.02
… 41
-
50 yrs
0.33
0.33
0
… 51
-
60 yrs
0.27
0.23
0.04**
… above 60 yrs
0.15
0.16
-
0.01
Head's Edu Level (yrs of schooling)
$
8.86
6.63
2.24***
Head Edu: Illiterate (no sch
ooling)
0.13
0.28
-
0.15***
up to
Primary (1
-
4)
0.04
0.09
-
0.05***
up to
Middle (5
-
8)
0.24
0.27
-
0.03
up to
High Secondary (9
-
12)
0.4
0
0.3
0
0.10***
… Grad & PG (> 12)
0.2
0
0.07
0.13***
HH Size
$
5.06
5.29
-
0.23***
Ratio Dep (
children<6 to women
18
-
60
)
0.07
0.07
0
Ratio Dep (
children 6
-
14 to women 18
-
60
)
0.15
0.16
0
Ratio Dep
(
aged>65 to women 18
-
60
)
0.05
0.03
0.01***
14
Household's M
o
nthly
P
er
C
ap
Earnings
#
4.58
3.65
0.94***
Household's
Asset Index
Score
@
1.6
0
-
0.68
2.29***
Migrated (Residing
before 1950)
0.33
0.23
0.10***
… 1950
-
70
0.12
0.26
-
0.14***
… 1971
-
80
0.12
0.18
-
0.06***
… 1981
-
90
0.14
0.18
-
0.04***
… Post 1990
0.29
0.15
0.15***
HH reported
Neighborhood
Unsafe
0.28
0.24
0.04**
Observations 1304 3168 4472
Notes: Unweighted sample means. Excluding households without women aged 18-20 yrs.
$
Figures
represent mean values in number (not percent term)
#
In INR (in thousands),
CPI adjusted for IHDS2 till 2012 and NoPoor till 2014
Table 5 below shows that marital status and age of women across databases have very similar
patterns. Roughly similar percentages of women are illiterate and just literates across slums and
non-slum areas. Slums have disproportionately higher percentage of primary level (up to 4 years
of schooling) educated women, whereas in rest of urban areas more women have completed
higher secondary level (12 years of schooling). Average level of education is higher in non-slum
areas (about 8.36 years) than slum areas (about 7.46 years). A very comparable proportion (10-
12%) of women are currently enrolled for study. More women in slum areas reported poor
health, whereas more women in non-slum areas have migrated to the current urban cities from
rural or other urban places due to marriage or education or due to family migration.
Table 5: Sample Means for variables related to Women (18-60) (percent)
Variables
IHDS-2
(Non-slum)
(1)
NoPoor
(Slum)
(2)
Difference
(1) - (2)
Marital Status: Currently Married
0
.72
0.69
0.03*
Separated/Divorced/ Widowed
0.07
0.10
-
0.03***
Never Married
0.21
0.21
0
Woman's Age
(in years)
$
35.43
34.49
0.94**
… below 30 yrs
0.42
0.46
-
0.03*
… between 31
-
40
0.22
0.24
-
0.03*
… between 41
-
50
0.23
0.19
0.04***
betwe
en 51
-
60
0.13
0.11
0.02
Currently Enrolled
0.12
0.10
0.03**
Education (yrs of schooling) $
8.36
7.46
0.89*
…Illiterate
0.18
0.17
0.02
…Literate
0.08
0.06
0.02**
…Primary Educated
0.14
0.25
-
0.11***
…Middle Educated
0.20
0.17
0.03**
…Secondary Educated
0.22
0.12
0.10***
…Higher Secondary
0.11
0.10
0.01
…Graduate and above
0.03
0.02
0.02***
15
Health Poor
0.02
0.10
-
0.08***
Migrated
0.47
0.34
0.13***
Monthly per cap other members income
#
4.16
3.26
0.91***
Observations
21
14
53
30
74
44
Note: Unweighted sample means
$ Figures represent mean values in number (not percent term)
#
In thousand INR (absolute value not percent)
We have created some illustrations that depict unconditional relations between female labor
force participation rates and certain covariates, these are available in the appendix section.
6. The Correlates of FLFP – Estimation Results
We now present the estimation results of our probit model for the aggregated samples from
slums of three cities (NoPoor) and their counterparts (IHDS2). In Table 6, the coefficients
represent average marginal effects, showing the change in the probability of being in the labor
force associated with a unit change in the explanatory variable, however for categorical variables
this is the difference with respect to the reference category. For the modelling purpose we take a
gradual approach – first, we introduce women’s individual level variables (col. 1 & 4), followed
by household’s and its head’s characteristics along with city fixed effects (col. 2 & 5), and lastly
introduce per capita other household members income (col. 3 & 6) to complete our model
specification.15
Let us first investigate women’s individual level explanatory variables. Marital Status of women
is one important factor affecting the women’s decision of participation. Being currently married
comes with a heavy restriction, the likelihood of working consistently falls sharply (by 16.6 to 21
percentage points) compared to the unmarried females, and the effect is noticeably stronger in
urban slums than their non-slum counterparts in the three metro cities. This post-marriage
withdrawal from labor force is purely urban culture/ phenomenon and is in stark contrust to what
observed in rural India. Sarkar, Sahoo, and Klasen (2019), Panda (1999) have found similar
effect in urban India. However, for the widow/separated/divorced women we notice a different
pattern in slums and non-slum areas the chances of working for the W-S-D women vis-à-vis
the never-married ones is significantly less (about 5.5 pp) among slum residents, which is a
15 To be noted that in this model we used a cut off 240 hrs of work in the preceding one year for any individual to be
considered as working. We have also estimated a model where work done for any hours to be considered as
working, estimation results are available in the appendix Table 9. Results are quite consistent and similar.
16
manifestation of economic distress; but for the non-slum cohort there’s no significant difference
observed between them.
Women’s age has the expected effect and follows a concave pattern. Participation first increases
and then declines with age - the employment probability is higher for middle aged women and
lower for both younger and older women. The peak working ages in slums and non-slum areas
are somewhat close and stand at 39.6 and 43.2 years, respectively. This could be because the
poorer economic conditions of the households of slum women may force them to continue to
work for some more extra years on average before they start to withdraw.
Contemporary literature stresses on U-shaped relationship between education and FLFP in India,
both in rural and urban sector. Let us first look at the col. 1 & 4, i.e. when only individual factors
considered, education seems to have a very strong U-effect on participation in the cities,
indicating women with no education, and lower and higher levels of education are more likely to
work as compared to those with secondary (base category) level of education, (see the average
unconditional participation rates in Figure 1 in Appendix-A); while in the slums only the higher
level of education seems to have strong positive effect on the participation, also a small positive
effect for the literates, is still U-shaped. But after controlling for other household factors, in both
slum and non-slum areas, the effect looks flat up to secondary level, and the uptick starts
appearing significantly stronger way beyond the secondary education in both areas. That is,
education has very similar effect on LFP decision of women from both slum and non-slum areas,
however it is not quite surprising, as the slum and non-slum clusters coexist side by side within a
city and face similar labor market challenges, and the relationship looks more like J-shaped with
an elongated flat bottom, much different from the pattern observed in the aggregated urban
sector in contemporary literature (see, Klasen and Pieters 2015).
As a matter of fact, this flat FLFP rate for up to secondary educated women in the metro cities is
a cause of concern for policy makers, since much articulated social stigma postulate does not
hold good anymore. This rather shows immense rigidities and lack of graded opportunities for
female workers in the labor market in urban metro areas, which seems to have conditions
different from those in non-metro urban areas.
Further, as expected, when the woman is currently studying/enrolled, she is less likely to work,
we found stronger significant negative effect in slums (about 21 pp) than in non-slum areas
17
(about 16 pp), and this stronger effect in slums shows that aspirations run equally or even higher
in slums for those who have reached college education (at 18 years of age, a student is expected
to reach college in normal course) either to get a better job or better qualified husband or both
and thereby observe restrain to work while studying.
Poor health of women seems to reduce the probability working, this is only observable in slums
though. Women, who have migrated from other towns or rural side due to marriage or family
migration and reside in non-slum areas, are less likely join workforce, but the same is not true for
the slum residents. On one side, this contrasting phenomenon points towards the structural issues
prevailing in the urban job markets that women who are not so well educated but migrated from
smaller towns or rural areas find themselves incompatible, on the other hand, for the slum
dwelling women, work choice arises out of their poor economic conditions.
Let us now turn the household heads’ characteristics. In non-slum areas, women are more likely
to go for work from female-headed households, but that does not hold good in slums. Head’s
education level has a gradually stronger negative effect on the household women’s work choice
in slums, however in non-slum areas this negative effect is visible, although in a feeble way, for
the heads who are beyond middle level educated. Klasen and Pieters (2015) also found similar
effect of head’s education for urban women, they used this as proxy for households’ wealth, but
we interpret that heads with higher education level enjoys better earning capability which makes
living more affordable and thereby reduces the requirement of women to seek for job, which is a
sign of patriarchal tendency.
Next, we controlled head’s age with the expectation that the older heads to be more conservative
than their younger counterpart and thereby females coming from those households are less likely
to be in the labor force. However, what we get here is little unexpected: in slums, females from
households with head’s age between 40-60 years are having about 4 pp significantly higher
participation vis-à-vis the households with heads above 60 years; while in non-slum areas,
women from the households with younger heads below age 30 years are about 7 pp less likely to
participate. Apparently, this age bracket of couples (40-60) in slums when their children are
going for higher studies or even getting married, which requires extra money to smooth out
higher expenses, perhaps this drives the women in slums to go out and work.
18
Coming to the other household level factors, we find a small linear negative effect of household
size among the slum households, but not in non-slum metro areas (this observation is not
consistent with other pan urban studies). Turning to the social groups, compared to the Hindu
Upper Castes, Muslim women are consistently less likely to participate in the labor market in
both areas (6-8 pp), this negative effect is even stronger in slums, and increases in magnitude
after the inclusion of other members income quartiles. Unlike many other studies on entire urban
India, we do not find any effect of caste (i.e. no significant difference among Hindu Upper Caste,
OBC and SC/ST) in non-slum areas of three metro cities; but coming to the slums women from
SC/ST and other religious minority households are more likely to participate, about 4 pp and 9.6
pp respectively. The ratio dependency variable for child (below 5 yrs) shows a negative impact
on FLFP; if the ratio were to decrease from 1 to 0.9 then the FLFP would increase by 0.046 pp in
slums and 0.038 pp in non-slum areas; the lower effect in non-slum areas might be due to better
affordability of those households to send their kids to crèches and playschools than the slum
households. We do not find any significant effect of other two ratios in any places.
For the migration/residency duration of families, we do not find any systematic pattern of effects
on FLFP in slums; however, women from families migrated during 1970-90 in non-slum areas
are 3 pp more likely to work than those who are residing before 1970. This maybe a mere co-
incidence that we could obtain this; when used any other time bracket or migration duration as
continuous variable, we could not obtain in pattern or effect at the aggregate metro levels
samples. This implies that migration duration of families either may have more micro level
patterns or may not have any systematic effect on its females, what may matter more the
individual female’s migration status.
It is perplexing to note that household’s report of neighborhood unsafe does not have any bite on
FLFP in the urban areas.
We do not find any effect of wealth (either before (see, col.2) or after the inclusion of other
household members income (see, col.3)) for the non-slum resident; but in the slums there exists a
negative effect of wealth initially (col.5), which disappears after the inclusion of other member
income (col.3). We find a significant negative income effect both in slum and non-slum areas –
effect gets stronger gradually with every successive higher quartile, in slums it is even stronger.
19
This implies that as the other household members income rises, women are less likely to go for
work indicating the income effect, for both slum and non-slum areas of 3 metros.
Table 6: Estimation Results (Average Marginal Effects) for
Slum & Non-slum Areas of Three Metro Cities
Pr (Working 240h/yr)
IHDS
2 (
Non
-
slum)
NoPoor
(Slum)
(1)
(2)
(3)
(4)
(5)
(6)
Marital Status (Ref: Never Married)
… Currently married
-
0.243***
-
0.190***
-
0.166***
-
0.271***
-
0.246***
-
0.212***
… wido
w/separate/divorce
0.052
0.0227
0.0205
-
0.0455
-
0.0650*
-
0.0548*
Woman's age
0.0384***
0.0366***
0.0386***
0.0433***
0.0395***
0.0370***
Sq.
woman's age
-0.000483*** -0.000464*** -0.000487*** -0.000515*** -0.000458*** -0.000428***
Woman's Edu (Ref: Seco
ndary)
… Illiterate
0.0923***
0.0508**
0.0427*
0.0238
-
0.0114
-
0.0241
… Literate
0.103**
0.0628
0.0461
0.0420**
0.018
0.0107
… Primary
0.0644**
0.0398
0.032
0.00674
-
0.0191
-
0.0195
… Middle
0.0326
0.0181
0.00964
0.000691
-
0.00741
-
0.0077
… Higher
Secondary
0.0493**
0.0581**
0.0602**
0.0174
0.0387*
0.0524**
… College/Grad
0.147***
0.167***
0.178***
0.149***
0.187***
0.210***
… PG & others
0.261***
0.264***
0.300***
0.219***
0.278***
0.317***
Woman currently Studying
-
0.158***
-
0.171***
-
0.159***
-
0.199***
-
0.211***
-
0.207***
Woman's Health
Poor
-
0.0129
-
0.0276
-
0.0291
-
0.0463**
-
0.0522***
-
0.0487***
Woman migrated
-
0.0138
-
0.0242
-
0.0337*
0.0274**
0.0102
0.0167
HH Head: Female
0.0453*
0.0426*
0.0219
-
0.00234
Head's Edu (Ref: Illiterate)
P
rimary
-
0.0596
-
0.0534
-
0.00597
-
0.00434
… Middle
0.00817
0.0136
-
0.0534***
-
0.0474***
… Higher Secondary
-
0.0597**
-
0.0512*
-
0.0657***
-
0.0575***
… College/Grad
-
0.0673*
-
0.043
*
-
0.0912***
-
0.0838***
Head's age cat: (Ref: >60yrs)
51
-
60 yrs
0.0276
0.0302
0.0304
0.0416**
… 41
-
50 yrs
0.0157
0.00556
0.0323*
0.0443**
… 31
-
40 yrs
0.0193
0.000959
0.0235
0.0263
up to
30 yrs
-
0.0592
-
0.0697*
0.051
0.0515
HH Social
group (
Ref: Hindu Upper Caste)
… Hindu OBC
0.0392*
0.035
6
0.026
0.0245
… All SC/ ST
-
0.0163
-
0.0142
0.0505***
0.0426***
… Muslim
-
0.0571**
-
0.0638***
-
0.0747***
-
0.0799***
… Other religions
0.0261
0.0366
0.0888**
0.0957**
Household
Size
0.00817
-
0.00318
-
0.00976
-
0.0211**
Sq
.
H
ousehold
Size
-
0.00
149
-
0.000924
0.000235
0.000724
Ratio dep. (Child <6 to women 18
-
60)
-
0.0370*
-
0.0376*
-
0.0337**
-
0.0456***
Ratio dep. (Child 6
-
14 to women 18
-
60)
0.00382
-
0.00326
0.0205**
0.00545
Ratio dep. (Old>64 to women 18
-
60)
0.0129
0.00188
-
0.0114
-
0.00897
HH Migrated (Ref:
Before 1970)
… migrated 1971
-
90
0.0257
0.0310*
-
0.00225
-
0.00225
… migrated after 1990
-
0.00148
-
0.0017
0.00687
0.00419
HH reported
neighborhood
unsafe
0.0131
0.00757
0.014
0.0156
City Fixed Effect (Ref: Mumbai)
… Kolkata
-
0.0105
-
0.0232
-
0.0291*
-
0.0616***
… Delhi
0.0364*
0.0078
-
0.0214
-
0.0366**
HH Asset Score (
Ref: Quartile
1)
20
… Quartile 2
-
0.036
-
0.0229
-
0.0369**
-
0.0201
… Quartile 3
-
0.0365
0.00138
-
0.0473***
-
0.00379
… Quartile 4
-
0.0383
0.0114
-
0.0
801***
-
0.00729
Per Cap Other Mem Inc (
Ref: Quartile
1)
… Inc quartile 2
-
0.0645***
-
0.143***
… Inc quartile 3
-
0.136***
-
0.198***
… Inc quartile 4
-
0.160***
-
0.242***
Observations
2114
2114
2114
5330
5330
5330
Pseudo
R
-
squared
0.133
0.167
0.193
0.085
0.116
0.151
AIC
1644.5
1623.7
1581.4
5279.3
5158.3
4962.1
* p<0.10, ** p<0.05, *** p<0.01
6.1. Estimation Results for Slums-clusters – a comparison three cities
Having discussed the covariates for aggregated slums and non-slum urban dwellers, let us now
turn to comparative analysis of the determinants of women’s LFP for slum-clusters at the city
level, this will uncover many micro regional characteristics. The Nopoor database captures
certain extra pieces of information, such as migrants’ state of origin, the tenancy types, where
they live (slums or rehabilitated colonies of Delhi and Mumbai), differences in legal protection
status of slums (available only for Mumbai). While a paper by Bag and Seth (2017) shows that
many of these factors do significantly affect their both monetary and non-monetary standard of
living; do they also affect, besides other factors already considered, the probability of working
for the women living in slums? For this purpose, we augment our original model with these
factors, estimation results are provided in the Table 7 below at the city level.
Across the city-slums, there is considerable differences in the negative likelihood of working of
currently married women vis-à-vis unmarried ones in Kolkata the effect is lowest (about 17
pp), while Delhi has the strongest negative impact (about 27 pp). In Northern India, people are
known to hold more conservative values about married women’s work choice, and Delhi’s case
seems to reflect that mindset. Similar factors are likely to affect the widows, separated and
divorced women from Delhi.
We observe a non-linear effect of age on participation in all cities, however the peak age differs
considerably – while in Delhi and Kolkata peak age is very similar, 46.8 and 50.8 years
respectively, but in Mumbai it is much lower and stands at 39 years. Perhaps this contrasting
picture has much to do with differences in economic standing of the households from slums of
three cities, worth noting that Bag and Seth (2017) has highlighted that on average slum dwellers
21
of Mumbai have much higher per capita income and multi-dimensional standard of living score
than the Delhi and Kolkata (these latter two have very similar standing on both counts).
Educational levels of women have very similar effect on FLFP in slums of three cities,
significant positive effects observed only for college or university attended women, (however in
Kolkata higher-secondary education also has a positive effect). Like our aggregated analysis, the
slums-clusters of each individual city also has a J-shaped relation between education level and
work participation for women. Being enrolled reduces the probability of working lowest and
strongest effects are observed in Kolkata and Mumbai respectively.
With respect to head’s education level, a very systematic pattern is observed in Mumbai, every
successive higher education level of head has a stronger negative effect on FLFP. Strikingly
different in Kolkata, women from households with higher secondary educated heads have 6.5 pp
higher chances of working. This is perhaps due to cultural differences between the cities. Head’s
age has an inverted U-shaped relation with FLFP only in Kolkata, however in other two cities
there’s no such bearing.
Does caste or religion matter at the slum level? In the previous section for aggregated analysis
we found that in slums when compared Hindu Upper Caste households, women from Muslim
households are less likely to join workforce, while for women from SC/ST households
probability is higher. But coming to the city level analysis, we find that only women from
Muslim households are less likely to work in all cities, while positive effect for SC/ST women
disappears.
Size of household only matters in Kolkata, in a non-linear way. Below 6 child ratio has a
significant negative effect on FLFP in Delhi and Mumbai, but not in Kolkata. Wealth effect is
only visible among the slumdwellers of Kolkata, FLFP falls as we move to second quartile, then
remains more or less flat. Monthly per capita other household members income (CPI adjusted)
shows a strong non-linear negative effect (convex) on FLFP in all cities. However again there
exists considerable difference in this income effect across city-slums – in Kolkata slums we
observe strongest negative effect.
Let us now turn to newly introduced variables. Tenancy type household does not seem to affect
work choice of women across cities, except for one case – women from households in Delhi that
22
are living in houses without any proper arrangements, they are most vulnerable ones, and
participation probability is about 16 percentage points higher than others.
Next, we look at the migration/residency duration and state of origin of the migrants. For
duration of migration, we only find a systematic pattern in Kolkata, where the women from
newer migrant households have higher probability of joining workforce. However, in Delhi,
women from households that migrated between 1950-80, have higher probability than those who
have residing on or before 1950. In Mumbai, residency/ migration duration of household does
not have any effect. Incidentally, the migration status of women at her individual level matters in
Mumbai (not in other two cities), those women who themselves have migrated (especially on
account of marriage) have about 6 pp higher probability of working. So, there is some link
between migration and likelihood of women’s work decision, but it has more localized flavor.
Is there any effect of migrant’s state of origin, which is proxy for regional/ethnic culture or
conservativeness embedded within the family? Bag, Seth and Gupta (2019) notes that in slums of
Kolkata about 33% households are from native state of West Bengal, while in Mumbai it’s 43%
and in Delhi only 2.3%, therefore rest of the slum dwellers are migrant from various other states.
Depending upon the concentration of these migrants, we divide them into natives and from other
regions. Six such regions are constructed.16 An interesting pattern is observed, women from the
migrant households from North Central regions of India are about 10 pp less likely to participate
compared to state-natives of Mumbai and Kolkata. And Delhi’s proximity to these states of that
region reflected in no significant difference between natives and the migrants from North Central
regions. This is an outcome of conservative mindset of the people from Northern region.
And finally, does residing in a rehabilitated or relocation colony or housing affect the work
choice? In Delhi we find that women from households residing in these rehabilitated colonies are
less likely to work (about 7.3 pp) than their regular slum counterparts, this is perhaps due
differences in living standards. To be noted that according Bag, Seth and Gupta (2019),
households residing in relocation and rehabitated colonies of Delhi have higher per capita
16 The six geographical regions consist of the following states –
North-Central: Uttar Pradesh, Bihar, Jharkhand and Uttarakhand; North-Western: Rajasthan, Haryana, Punjab,
Himachal Pradesh, Jammu and Kashmir and Delhi; Central: Madhya Pradesh, Chhattisgarh;
Western: Maharashtra, Gujarat, Goa, and Daman-Diu; Southern: Tamil Nadu, Kerala, Karnataka, Andhra and
Telengana; Eastern: West Bengal and Orissa.
The native state of Kolkata, and Mumbai are West Bengal, Maharashtra, respectively; but for Delhi we bring slight
variation it’s native to Delhi plus people from North-Western states.
23
income and higher multi-dimensional achievements than those lving in Jhuggi-Jhopri slum
clusters. In Mumbai, however, we get a mixed picture, but clearly women from SRA housing
are less likely to participate in workforce than those living in slums.
Table 7: Estimation Results (Average Marginal Effects) for City Wise Slums
Pr
(Working
)
(def
:
240
h
r
/yr
)
Delhi
Kolkata
Mumbai
Marital Status (Ref: Never Married)
… Currently married
-
0.265***
-
0.173**
*
-
0.222***
… Widowed/Separated/Divorced
-
0.124**
-
0.0341
-
0.0182
Woman's age
0.0397***
0.0253***
0.0445***
Sq.
woman's age
-
0.000428***
-
0.000249***
-
0.000571***
Woman's Edu (Ref: Secondary)
… Illiterate
-
0.0177
0.0386
-
0.0388
… Literate
0.042
0.
0145
0.0125
… Primary
0.0116
0.0515
-
0.0673*
… Middle
0.0249
0.0227
-
0.0281
… Higher Secondary
0.01
0.188***
0.0373
… College/Grad
0.209***
0.310***
0.206***
… PG & others
0.231**
0.455***
0.378***
Woman currently Studying
-
0.216***
-
0.137***
-
0.26
1***
Woman's Health Poor
-
0.0857***
-
0.0418
-
0.0286
Woman migrated
-
0.000965
-
0.0111
0.0581***
HH Head: Female
0.0149
0.0433
-
0.0371
Head's Edu (Ref: Illiterate)
Primary
-
0.0628*
0.0336
-
0.00889
… Middle
-
0.0298
0.00296
-
0.0876***
… Higher Seco
ndary
-
0.0457*
0.0648*
-
0.0978***
… College/Grad
0.0398
-
0.0541
-
0.148***
Head's age cat: (Ref: >60yrs)
… 51
-
60 yrs
0.0144
0.0785**
0.0324
… 41
-
50 yrs
0.0247
0.109***
0.0161
… 31
-
40 yrs
0.0452
0.103**
-
0.0279
up to
30 yrs
0.0635
0.09
0.000552
H
H Social
group (
Ref: All SC/ ST)
… Hindu Upper Caste
-
0.0396
-
0.0445
-
0.0154
… Hindu OBC
-
0.0178
0.0215
-
0.000993
Muslim OBC
-
0.110***
-
0.0456
-
0.160***
… Muslim general
-
0.105***
-
0.0637*
-
0.0926***
… Other religions
-
0.0346
0.0487
0.115*
Hou
sehold
Size
0.000506
-
0.0346**
-
0.0172
Sq
.
Household
Size
-
0.000195
0.00186**
0.00034
Ratio dep. (Child <6 to women
18
-
60)
-
0.0508**
0.0218
-
0.0636**
Ratio dep. (Child 6
-
14 to women 18
-
60)
0.0156
-
0.013
0.0141
Ratio dep. (Old>64 to women 18
-
60)
-
0.049
8
0.0249
-
0.0036
HH reported
Neighborhood
unsafe
0.0234
-
0.00798
0.0251
HH Asset Score (Ref: Quartile 1)
Quartile 2
0.00273
-
0.0865***
-
0.00902
… Quartile 3
0.00379
-
0.0666**
0.0186
… Quartile 4
0.0437
-
0.0687*
0.042
M
o
n
thly
per cap other mem in
c
ome
(
‘000
)
-
0.0430***
-
0.0687***
-
0.0496***
Sq
.
mn_
per_cap_
oth_mem_inc
0.00171***
0.00196***
0.00100***
24
Tenancy Type (Ref: Owned House)
… Rented (Thika/Pagdi)
-
0.0436
-
0.0438
-
0.000655
… Rented (Informal)
0.0108
-
0.0463
0.0506
… Rented (Other arra
ngements)
0.160*
-
0.0309
0.0232
HH Migrated (Ref: Before 1950)
… between 1951
-
80
0.0508*
0.0578**
-
0.0265
… post 1980
0.0332
0.118***
-
0.0386
Migrant HH (Ref: Native to State …)
Delhi & North West
West Bengal
Maharas
h
tra
… from North Central states
0.0197
-
0.106***
-
0.106***
… from Western States
-
-
-
0.00125
… from
Southern states
-
-
0.0000157
… from Foreign
-
-
0.00246
-
0.0371
… Rests
0.0968***
-
0.209***
-
Slum Type
Delhi (
Ref: Jhuggi Jhopdi)
… Resettled & Rehabilitated
Colony
-
0.0731***
Slum Type Mumbai (Ref: Protected till 1998)
Declared/Notified Slums (post 1998)
-
0.0372*
… Non
-
notified/ Protection status not available
-
0.0373
… Rehabilitated (SRA)
-
0.0590*
Observations 2026 1296 1976
Pseudo R
-
squared
0.15
0.211
0.204
AIC
1917.6
1177.5
1852.3
FLFP
(def
240h
/
yr)
22.07%
22.01%
24.46%
* p<0.10, ** p<0.05, *** p<0.01
7. Conclusion
According to IMF chief Christine Lagarde, 217 million women are missing from the Indian labor
force, and if women and men were equally represented, it would boost India’s economy by 27
percent. Urban sector is the one which lags the most. The very low and stagnant female labor
force participation in the urban sector drew attention of many researchers, but the focus remained
on the entire urban sector. However, within urban sector there has been an ever-growing cluster
of slums and its population, but there exists no study which covered the females from these
slums – factors that affect their decision to work, the type of work they do, the constraints they
face, how these factors are similar or different from their other urban counterparts. Should policy
makers and researchers interested in devising policy to boost participation rate in urban areas, it
is important to be apprised of the process of decision making within the households in slum and
non-slum areas. Main constraint is lack of data, NoPoor survey in slums of three metro cities was
an important step in that direction.
This paper is the first to explicitly account for female workforce participation and its
determinants for the slum dwellers of three metro cities in India and fills a critical gap in
25
literature. It also brings out a comparative picture of similarities and differences in determinants
of workforce participation in aggregated slums and non-slum areas of the three metro cities, it
goes beyond in identifying important determinants of LFP of women from the slum-clusters
across three metro cities.
Our study shows that the workforce participation rates for slum dwelling women (aged 18-60)
from three metro cities are higher than the non-slum urban counterparts, but surprisingly the
difference is not vastly high; in fact NoPoor estimates of LFP rates are quite comparable to that
of NSSO (as per UPSS definition) for the urban sector, but IHDS-2 estimates are tad lower.
We find that single women are much more likely to work, whereas severe restrictions that comes
with marriage. Presence of children below 5 in the family consistently affects women’s
participation throughout, indicating that provision of affordable childcare facilities in localities or
in workplace can be an important policy instrument.
We find a flat bottom J-shape relation between women’s education and work choice that remains
consistent between slums and non-slum urban areas of three metro cities, as well as across-city
slum-clusters. This points towards rigid urban metro labor market and lack of opportunities for
women particularly with low or moderate education. Additionally, we find that an increase in
earning of other household members leads to lower working probabilities of women from both
slums and non-slum areas, moreover this income effects renders the wealth effects void. This
instance supports so-called ‘added worker effect’, women mostly participate in the laborforce in
order to smooth household consuption. In fact, this income effect of other members works in a
stronger way (across successive quartiles) in slums than non-slum areas, and that is perhaps the
most important reason why slum women’s LFPR is not vastly higher than that of women from
over-all three metro cities. Further, household-heads with more education often restrict the
women to work, and happens more strongly in slums. This finding is consistent with the
literature which highlights the possibility that women may be – themselves disinclined to be
employed because of family’s economic and social status, or even worse, compelled not to take
up paid jobs or even quit with the enhancement of family’s economic status,. Moreover, we show
that caste does matter in slums of three metro cities in the expected inverse hierarchical way, but
surprisingly not in over-all metro urban areas (unlike many studies). Muslim women consistently
show lower participation across the board.
26
Women in urban areas, both from slum and non-slum areas, continue to experience a range of
multiple challenges from the inflexible and unreformed labour market relating to access to
employment, choice of work, as well as poor working conditions, employment security and wage
parity, moreover they also face discrimination at home, while required to balance the competing
burdens of work and family responsibilities. While some of the factors affecting female
participation, e.g. socio-religious norms, patriarchal attitudes, overall labor market conditions, or
even level of female education, cannot immediately be affected by policy reforms, however
some others can. Beyond standard labor force participation rates, policymakers should pay
attention to how women can access better jobs and take advantage of new labor market
opportunities that arise as a country grows. Therefore, policies should be designed to tackle both
supply- and demand-side dimensions, including access to better education and training programs
and access to childcare, access to formal (micro) finance, as well as other supportive institutions
and legal measures to ease the burden of domestic duties, enhance women’s safety, and
encourage the absorption women in manufacturing and especially in low-level services, that has
been seen as the gateway for women in other poor countries. Importance of positive effects of
sensatization activities by Women’s Self-help groups must be recognized and encouraged
further.17
Lastly, patriarchal tendencies and social norms often thrive on low female education and that in turn
affect the FLFP. Over the years, an emphasis was given to keep young girls in schools, however many
fear that all that extra years of schooling actually favoured the parental ploy to improve a daughter’s
prospects not in the labour market but in the (arranged-) marriage market (The Economist 2018).
Therefore, a further push is required by shifting the attention towards ensuring that they receive a good
quality education, and particularly beyond that crtical junior secondary level, and thus they are not only
able to bargain better with the family patriarchs or adverse societal norms, but also help them taking
advantage of training opportunities, that, in turn, will increase their chances of overcoming other
barriers to finding decent employment.
17 In a related paper (forthcoming) this author observed the positive influence of women’s membership/ association
with the self-help groups both in rural and urban sectors, using IHDS-2 data.
27
References
Agarwal, Siddharth. 2011. “The State of Urban Health in India; Comparing the Poorest Quartile
to the Rest of the Urban Population in Selected States and Cities.” Environment and
Urbanization.
Agrawal, Tushar. 2014. “Gender and Caste-Based Wage Discrimination in India: Some Recent
Evidence.” Journal for Labour Market Research.
Andres, Luis A et al. 2017. World Bank Policy Research Working Paper Precarious Drop
Reassessing Patterns of Female Labor Force Participation in India.
Bag, S., and S. Seth. 2017. “Does It Matter How We Assess Standard of Living? Evidence from
Indian Slums Comparing Monetary and Multidimensional Approaches.” Social Indicators
Research.
Bag, Sugata, Suman Seth, and Anish Gupta. 2016. “A Comparative Study of Living Conditions
in Slums of Three Metro Cities in India.” SSRN Electronic Journal.
———. 2019. Centre for Contemporary Indian Studies, University of Colombo Slums and
Slumdwellers in Three Metro Cities of India: A Comparative Study of Living Conditions.
Colombo.
Bhalla, Surjit S., and Ravinder Kaur. 2010. “Labour Force Participation of Women in India:
Some Facts, Some Queries.” Asia Research Centre, Working Paper 40, London School of
Economics: 1–26.
Carr-Hill, Roy. 2013. “Missing Millions and Measuring Development Progress.” World
Development.
Chatterjee, Esha, Sonalde Desai, and Reeve Vanneman. 2018. “Indian Paradox: Rising
Education, Declining Women’s Employment.” Demographic Research 38(1): 855–78.
Chatterjee, Urmila, Rinku Murgai, and Martin Rama. 2015. “Job Opportunities along the Rural-
Urban Gradation and Female Labor Force Participation in India.” (September): 37.
Chen, Marty, and Jean Dreze. 1992. “Widows and Health in Rural North India.” Economic and
Political Weekly.
Das, Maitreyi Bordia, and Sonalde Desai. 2003. “Why Are Educated Women Less Likely to Be
Employed in India?: Testing Competing Hypotheses.” Social Protection Discussion Paper
Series,Human Development Network, The World Bank (0313).
http://www.sonaldedesai.org/das_and_desai_2003_on_women.pdf.
28
Das, Maitreyi Bordia, and Ieva Žumbytė. 2017. “The Motherhood Penalty and Female
Employment in Urban India.” (March).
https://openknowledge.worldbank.org/bitstream/handle/10986/26347/WPS8004.pdf?sequen
ce=1.
Das, Sonali, Sonali Jain-chandra, Kalpana Kochhar, and Naresh Kumar. 2015. “Women Workers
in India: Why So Few Among So Many?” IMF working Paper 55.
Delhi, Economic survey of. 2018. “Economic Survey 2018-19.” Textbook of family medicine.
Desai, S., and D. Jain. 1994. “Maternal Employment and Changes in Family Dynamics: The
Social Context of Women’s Work in Rural South India.” Population & Development
Review 20(1): 115–36.
Dupont, Véronique. 2008. “Slum Demolitions in Delhi since the 1990s: An Appraisal.”
Economic and Political Weekly.
Eswaran, Mukesh, Bharat Ramaswami, and Wilima Wadhwa. 2012. “Status, Caste, and the Time
Allocation of Women in Rural India.” Economic Development and Cultural Change.
Ghosh, Jayati. 2009. “Migration and Gender Empowerment: Recent Trends and Emerging
Issues.” Human Development Research Paper (HDRP) Series.
Gonzales, Christian et al. 2015. “Catalyst for Change: Empowering Women and Tackling
Income Inequality.” Staff Discussion Notes.
Hirway, Indira, and Sunny Jose. 2011. “Understanding Women’s Work Using Time-Use
Statistics: The Case of India.” Feminist Economics.
Kanjilal-Bhaduri, Sanghamitra, and Francesco Pastore. 2018. “Returns to Education and Female
Participation Nexus: Evidence from India.” Indian Journal of Labour Economics.
Kapsos, Steven, Andrea Silberman, and Evangelia Bourmpoula. 2014. “Why Is Female Labour
Force Participation Declining so Sharply in India?” ILO Research Paper (Research Paper
No. 10): 1–51.
KILM-ILO. 2014. International Labour Organisation Key Indicators of the Labour Market.
http://ina.bnu.edu.cn/docs/20140604152414095087.pdf.
Kingdon, Geeta Gandhi, and Jeemol Unni. 2001. “Education and Women’s Labour Market
Outcomes in India.” Education Economics.
Klasen, Stephan, and Janneke Pieters. 2015. “What Explains the Stagnation of Female Labor
Force Participation in Urban India?” World Bank Economic Review 29(3): 449–78.
29
LEWIS, W. ARTHUR. 1954. “Economic Development with Unlimited Supplies of Labour.” The
Manchester School.
Nicholas, Ralph W., and M. N. Srinivas. 1967. “Social Change in Modern India.” Man.
Panda, P.K. 1999. Poverty and young women’s employment: Linkages in Kerala, Working Paper
(Thiruvananthapuram, Centre for Development Studies).
Psacharopoulos, George, and Zafiris Tzannatos. 1989. “Female Labor Force Participation: An
International Perspective.” World Bank Research Observer.
Rani, Uma, and Jeemol Unni. 2009. “Do Economic Reforms Influence Home-Based Work?
Evidence from India.” Feminist Economics.
Reddy, D Narasimha. 1979. “Female Work Participation in India: Facts, Problems, and Policies.”
Indian Journal of Industrial Relations.
Rustagi, Preet. 2013. “Changing Patterns of Labour Force Participation and Employment of
Women in India.” Indian Journal of Labour Economics.
Sarkar, Sudipa, Soham Sahoo, and Stephan Klasen. 2019. “Employment Transitions of Women
in India: A Panel Analysis.” World Development.
Sudarshan, Ratna M, and Shrayana Bhattacharya. 2008. 2006 ILO Asia-Pacific Working Paper
Series Through the Magnifying Glass : Women ’ s Work and Labour Force Participation in
Urban Delhi Subregional Office for South Asia , New Delhi Through the Magnifying Glass :
Women ’ s Work and Labour Force Participati.
———. 2009. “Through the Magnifying Glass : Women ’ s Work and Labour Force
Participation in Urban Delhi.” Economic & Political Weekly.
Swaminathan, Madhura, and Sharmila Majumdar. 2006. “Workers in Urban India: Changes in
Gender Segregation of Occupations, 1983-2000.” Indian Journal of Labour Economics.
Thomas, Jayan Jose. 2012. “India’s Labour Market during the 2000s Surveying the Changes.”
Economic and Political Weekly 47(51): 39–51.
30
APPENDIX – A: Figures
31
32
33
APPENDIX – B: Tables
Table 8: Asset Distribution Across databases
Items IHDS2 NoPoor Diff (1)- (2)
Bicycle
0.46
0.21
0.26***
Sewing machine
0.43
0.24
0.18***
Motorcycle
0.45
0.19
0.26***
Car/Auto
0.12
0.02
0.10***
Telephone
0.14
0.04
0.11***
Cellphone
0.95
0.96
0
TV
0.96
0.91
0.06***
Fr
idge
0.75
0.52
0.23***
Fan
1.00
0.97
0.03***
AC/cooler
0.72
0.28
0.44***
Washing machine
0.33
0.20
0.13***
Computer/Laptop
0.21
0.13
0.08***
Chair_table
0.93
0.59
0.34***
Cot
0.87
0.79
0.09***
Improved floor
0.98
0.94
0.03***
Improved wall
0.97
0.9
2
0.06***
Improved roof
0.88
0.42
0.46***
Electric
i
ty
1.00
0.99
0
LPG
0.95
0.82
0.13***
Separate
kitchen
0.69
0.33
0.36***
Personal
toilet
0.88
0.30
0.58***
Piped water at premise
0.70
0.41
0.29***
Type of housing
Pu
cca
0.78
0.39
0.38***
S
emi_
P
ucca
0.21
0.59
-
0.38***
Kutcha
/
Other
0.01
0.02
-
0.01
Observations
1304
3168
4472
Table 9: Estimation Results (Average Marginal Effects)
Pr (Working any hr pa)) IHDS2 NoPoor
(1)
(2)
(3)
(4)
(5)
(6)
Marital Status (Ref: Never
Married)
… Currently married
-
0.250***
-
0.201***
-
0.175***
-
0.290***
-
0.265***
-
0.229***
… Widow/separated/divorced
0.0382
0.0126
0.0128
-
0.0583*
-
0.0830**
-
0.0730**
Woman's age
0.0380***
0.0374***
0.0393***
0.0445***
0.0404***
0.0380***
Sq.
woma
n's age
-0.000475*** -0.000471*** -0.000494*** -0.000528*** -0.000468*** -0.000438***
Woman's Edu (Ref: Secondary)
34
… Illiterate
0.0904***
0.0411
0.033
0.0216
-
0.0144
-
0.0288
Literate
0.0956**
0.052
0.0361
0.0485**
0.0227
0.0135
… Primary
0.0688*
*
0.039
0.0315
0.00252
-
0.0284
-
0.0296
… Middle
0.0289
0.0123
0.00368
0.00437
-
0.00547
-
0.00617
… Higher Secondary
0.0469**
0.0567**
0.0596**
0.0133
0.036
0.0511**
College/Grad
0.148***
0.170***
0.181***
0.136***
0.176***
0.201***
… PG & others
0.255
***
0.258***
0.294***
0.219***
0.279***
0.322***
Woman currently Studying
-
0.165***
-
0.177***
-
0.166***
-
0.221***
-
0.233***
-
0.227***
Woman's Health Poor
-
0.019
-
0.0333
-
0.0345
-
0.0537***
-
0.0622***
-
0.0586***
Woman migrated
-
0.0116
-
0.0201
-
0.0302*
0.0
220*
0.00948
0.0165
HH Head: Female
0.0421*
0.0396
0.0299*
0.00433
Head's Edu (Ref: Illiterate)
… primary
-
0.0699*
-
0.0629*
-
0.000483
0.000531
Middle
0.00631
0.013
-
0.0421**
-
0.0363**
… Higher Secondary
-
0.0627**
-
0.0521*
-
0.0610***
-
0.
0527***
… College/Grad
-
0.0751**
-
0.0484
-
0.0906***
-
0.0821***
Head's age cat: (Ref: >60yrs)
… 51
-
60 yrs
0.0286
0.029
0.0364*
0.0492**
41
-
50 yrs
0.00798
-
0.00471
0.0334*
0.0461**
… 31
-
40 yrs
0.00781
-
0.0129
0.0289
0.0323
up to
30 yrs
-
0.0698
-
0.0836*
0.0518
0.0521
HH Social
group (
Ref: Hindu Upper Caste)
… Hindu OBC
0.0381
0.0348
0.0266
0.0249
… All SC/ ST
-
0.0173
-
0.0145
0.0487***
0.0399***
… Muslim
-
0.0641***
-
0.0707***
-
0.0687***
-
0.0742***
… Other religions
0.021
6
0.0317
0.0949**
0.102**
H
ousehold
_Size
0.0113
0.000216
-
0.0127
-
0.0251***
Sq
.
H
ousehold
Size
-
0.00165
-
0.00111
0.000387
0.000921
Ratio dep. (Child <6 to women 18
-
60)
-
0.0322
-
0.0334*
-
0.0336**
-
0.0462***
Ratio dep. (Child 6
-
14 to women 18
-
60)
0
.0036
-
0.00377
0.0220***
0.00575
Ratio dep. (Old>64 to women 18
-
60)
0.0106
-
0.00207
-
0.00448
-
0.00191
HH Migrated (Ref:
After 1990)
… migrated 1971
-
90
0.0174
0.0229
-
0.00596
-
0.00538
… migrated 1970 or before
0
0
0
0
HH reported neighborhoo
d unsafe
0.0135
0.00769
0.0129
0.0144
City Cat (Ref: Mumbai)
… Kolkata
-
0.0102
-
0.0214
-
0.019
-
0.0541***
… Delhi
0.0491**
0.022
-
0.0175
-
0.0337**
HH Asset Score (Ref: Quartile 1)
… Quartile 2
-
0.0445*
-
0.0314
-
0.0456***
-
0.0267
… Qu
artile 3
-
0.0398
-
0.00295
-
0.0597***
-
0.0118
… Quartile 4
-
0.0495*
-
0.000643
-
0.0936***
-
0.0139
Per Cap Other Mem Inc (Ref: Quartile 1)
… Inc quartile 2
-
0.0643***
-
0.156***
… Inc quartile 3
-
0.132***
-
0.212***
… Inc quartile 4
-
0.1
62***
-
0.265***
Observations 2114 2114 2114 5330 5330 5330
Pseudo R
-
squared
0.133
0.167
0.193
0.085
0.116
0.151
AIC
1644.5
1623.7
1581.4
5279.3
5158.3
4962.1
* p<0.10, ** p<0.05, *** p<0.01
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
In this paper, we aim to understand whether low labour market returns to education in India are responsible for low female work participation. The National Sample Survey Office—Employment Unemployment Survey unit level data of India for the year 2011–2012 is used to examine the relationship between educational attainment and labour market participation through gender lens. Results show that women’s education has a U-shaped relationship with paid work participation. The probability to participate in the paid labour market increases with education levels higher than compulsory secondary schooling. The labour market returns to education are insignificant and low for lower levels of education, increasing significantly along the educational levels. Technical education equivalent to degree level or above has high returns for men and women. However, women with technical education have very low levels of participation. Vocational training also provides a positive return. Our results suggest that to increase participation, women need to be educated above secondary level and receive broader technical education and more vocational training.
Article
Full-text available
BACKGROUND Theories of human capital would suggest that with more education, women acquire greater skills and their earnings increase, resulting in higher labor force participation. However, it has been long known that in India, women's education has a U-shaped relationship with labor force participation. Part of the decline at moderate levels of education may be due to an income effect whereby women with more education marry into richer families that enable them to withdraw from the labor force. OBJECTIVE The paper uses the first comprehensive Indian income data to evaluate whether the other family income effect explains the negative relationship between moderate women's education and their labor force participation. METHODS Using two waves of the India Human Development Survey, a comprehensive measure of labor force participation is regressed on educational levels for currently married women aged 25-59. RESULTS We find a strong other family income effect that explains some but not all of the Ushape education relationship. Further analyses suggest the importance of a lack of suitable employment opportunities for moderately educated women. CONCLUSION Other factors need to be identified to explain the paradoxical U-shape relationship. We suggest the importance of occupational sex segregation, which excludes moderately educated Indian women from clerical and sales jobs.
Article
Full-text available
As part of Sustainable Development Goals, the United Nations have set targets of upgrading slums and reducing poverty in all its dimensions by 2030. Policies towards improving the living conditions of slum-dwellers require proper assessment of their standard of living as well as understanding the associated characteristics. In this paper, using slum-level primary household survey data from three largest Indian cities, we, first, assess the standard of living of slum dwellers using both monetary and non-monetary approaches and then explore how various household and spatial characteristics are consistently or differently associated with both forms of assessments. We use standard monetary indicators, but to assess non-monetary standard of living, use a counting approach framework and justify the selection of specific indicators in the context of slums. Our analysis yields some interesting observations as some characteristics are differently associated with monetary and non-monetary living standards, which should affect policy designs in slums.
Article
Full-text available
This paper uses successive rounds of National Sample Survey Organization data from 1993-94 to 2011-12, and draws from census data. This paper (i) provides a description of nearly two decades of patterns and trends in female labor force participation in India; (ii) estimates the extent of the recent decline in female labor force participation; and (iii) examines and assesses the contribution of various demographic and socioeconomic factors in explaining the female labor force participation decision and the recent the drop. The analysis finds that female labor force participation dropped by 19.6 million women from 2004-05 to 2011-12. Participation declined by 11.4 percent, from 42.6 to 31.2 percent during 1993-94 to 2011-12. Approximately 53 percent of this drop occurred in rural India, among those ages 15 to 24 years. Factors such as educational attainment, socioeconomic status, and household composition largely contributed to the drop, although their effects were more pronounced in rural areas. Specifically, the analysis finds a U-shaped relationship between levels of educational attainment and female labor force participation. The decomposition of the contribution of these various determinants to the female labor force participation decision suggests that stability in family income, as indicated by the increasing share of regular wage earners and declining share of casual labor in the composition of family labor supply, has led female family members to choose dropping out of, rather than joining, the labor force. The findings of this paper suggest that conventional approaches to increasing female labor force participation (such as education and skills and legal provisions) will be insufficient. Policies should center on promoting the acceptability of female employment and investing in growing economic sectors that are more attractive for female employment.
Article
Full-text available
Since the 1990s, India has seen robust economic growth, rising wages, steady fertility decline, increased urbanization, and expanded educational attainment for males and females. But unlike other countries that have undergone similar transitions, urban women's employment has refused to budge, never crossing the 25 percent mark. This paper fills a critical gap in policy research on women's employment in India. The discussion is situated in the normative construction of motherhood and the gendered nature of caregiving in India. The analysis uses pooled data from six rounds of the National Sample Surveys to examine the effects of having a young child on mothers' employment in urban India over 1983-2011. The analysis also looks at household structure, and analyzes the effects of other household members on women's labor supply. The results show that although the onus of childbearing may have reduced, that of caregiving has increased. Having a young child in the home depresses mothers' employment, an inverse relationship that has intensified over time. Further, living in a household with older children and women over the age of 50 is positively associated with women's employment. These results show that the care of young children is an increasingly important issue in women's employment decisions, in a context where formal childcare is practically nonexistent. These results have significant implications for policy to raise women’s labor force participation in India.
Article
This study analyses employment transitions of working-age women in India. The puzzling issue of low labour force participation despite substantial economic growth, strong fertility decline and expanding female education in India has been studied in the recent literature. However, no study so far has looked into the dynamics of employment in terms of labour force entry and exit in this context. Using a nationally representative panel dataset, we show that women are not only participating less in the labour force, but also dropping out at an alarming rate. We estimate an endogenous switching model that corrects for selection bias due to initial employment and panel attrition, to investigate the determinants of women’s entry into and exit from employment. We find that an increase in wealth and income of other members of the household leads to lower entry and higher exit probabilities of women. Along with the effects of caste and religion, this result reveals the importance of cultural and economic factors in explaining the low workforce participation of women in India. We also explore other individual and household level determinants of women’s employment transitions. Moreover, we find that a large public workfare program significantly reduces women’s exit from the labour force. Our study indicates that women’s entry and exit decisions are not necessarily symmetric, and it is important to consider the inter-temporal dependence of labour supply decisions.
Article
The growth of gross domestic product in every sub-sector of the Indian economy accelerated during the second half of the 2000s, compared to the first half of the decade. However, employment growth in most sectors except construction decelerated. This jobless growth was partly the result of positive changes such as the reduction of "distress employment" in agriculture, created during the previous half-decade, and an expansion in the population of students. Rural wages rose and average educational levels of the workforce improved. Government interventions in rural India since the mid-2000s, particularly the Mahatma Gandhi National Rural Employment Guarantee Act, seemed to have aided these positive transformations. However, manufacturing employment in the country fell and employment growth slowed down in most constituents of the services sector. The new jobs generated were predominantly in rural construction. The slow progress in the diversification of India's employment structure has led to large-scale withdrawal of women from the labour force, with the number of women thus "missing" being as large as the population of Brazil.
Article
This study shows empirically that gender inequality and income inequality are strongly interlinked, even after controlling for standard drivers of income inequality. The study analyzes gender inequality by using and extending the United Nation’s Gender Inequality Index (GII) to cover two decades for almost 140 countries,. The main finding is that an increase in the GII from perfect gender equality to perfect inequality is associated with an almost 10 points higher net Gini coefficient. For advanced countries, with higher gender equity in opportunities, income inequality arises mainly through gender gaps in economic participation. For emerging market and developing countries, inequality of opportunity, in particular in education and health, appear to pose larger obstacles to income equality.