Gendered Choice of Paid and Unpaid Work in Sri Lanka: Issue of Female Labour Force Participation Revisited

The insignificance of the same variables in the labour force equation which were significant in determining unpaid work indicates that household care is indeed a hindrance for women to[r]

Gendered Choice of Paid and Unpaid Work in Sri Lanka: Issue of Female Labour Force Participation Revisited

Asia Convening 2019 – Vietnam

August 12 - August 14

Author – Dilshani N Ranawaka (Sri Lanka)

Gendered Choice of Paid and Unpaid Work in Sri Lanka: Issue of Female Labour Force Participation Revisited

Abstract

Given the free-education system in Sri Lanka, gender gap in literacy is very low (non-existent among youth) According to the Department of Census and Statistics (DCS, 2013/14), 62.7% of the university entrants are females However, the Labour Force Participation Rate (LFPR) among women in 2018 was 34% (DCS, 2018Q3) In spite of the gendered role as a mother and wife, and being the focal point of family and society, there is still a lacuna in the empirical literature that offers evidence that their gender identity is a constraint to them in entering the labour market in Sri Lanka While the literature identifies three broad categories of economic activities; regular jobs, self-employment and moonlighting, and routine work such as child care and household chores, the burden of unpaid work can be heavily weighed upon the shoulders of women in society This paper attempts to examine and identify if there exists gendered segregation among paid and unpaid work which could vary on domains such as education, marriage, sectoral divisions of rural and urban sectors and wealth The estimation of univariate and bivariate probit model using the Labour Force Survey (LFS) data in 2016, reveals that being a female increases the probability of selecting unpaid work by 52% as opposed to selecting paid work which decreases by a percentage

of 45% In spite of sociological dimensions explored, the study concludes that this gendered segregation is influential in discouraging women from entering the labour force in addition to other unobserved variables

Keywords:

Gender, female labour force participation, paid work, unpaid work, labour market

Author – Dilshani N Ranawaka

Reference Number - GE 04

Introduction

Gender inequality among certain characteristics such as education and health in Sri Lanka is reportedly minimum in comparison to other regional countries within the South Asian region Free education in the country is the predominant factor that influences these societal conditions leaving positive externalities evident in various dimensions Epidemiological transition from communicable diseases to non-communicable diseases is one such example The difference between the literacy rates among the two gender is as low as 2% The life expectancy of women

is above par compared to other competitive giants in the region

However, even amidst such positive looking prospects, the labour market conditions within the country still seems to be stagnant A thorough monitoring of the annual bulletin of the Labour Force Survey (LFS) reveals two alarming facts; the relatively low labour force participation rate and the higher unemployment rate in the labour market As of 2018, the statistics reveal that the labour force participation rate among females is as low as 33.6% compared to 73% of males within the same year The report also reveals that of the 34.9% economically active females, 7.1% are still unemployed The female representatives in the parliament is 6% in Sri Lanka while the regional static is 18% where this particular statistic is significantly lower in Sri Lanka This inactive labour force is one of the dominant factors why the SDG goal on gender inequality is stagnant at the moment

With the equal accessibilities granted for both genders, the lower labour force participation among women could be a reflection of an existing but an unexplored labour market constraint that prevents females from entering the labour market Endorsing this argument is supported by the significantly higher number of females (77%) contributing as a family worker as opposed to being

an own account worker, an employer or an employee which is dominated by male workers ranging from 66%-88% of the work force

While unemployment of the country has a very sound literature, the burden of family, marriage and care rested upon on the shoulders women and its economic impacts are areas that are overlooked This study hopes to identify if there exists societal norms and preassigned gender roles

as a constraint towards employment and labour force participation between males and females This question is mainly addressed through exploring the impact of sociological, household characteristics of an individual The main research question is addressed through looking at the following questions;

1 Is there a significant gendered segregation among paid and unpaid work?

2 What are the predominant micro-level, sociological characteristics that determine unpaid work prevalent women?

3 Is there a significant impact by unpaid work on the labour force participation in Sri

Lanka?

4 What are the salient sociological characteristics that binds women from entering the labour force in Sri Lanka?

The paper is structured as follows The first part consists of the literature review, followed up by the analytical framework, discussion of results and the conclusion

Literature review

Introduction

Labour supply is an area heavily studied by labour economists specially for the purpose of shaping sound policies in a country’s labour market In spite of predetermined gender roles found in every walk of life, gendered segregation of particular fields, gender discrimination and gender wage gap are some of the common topics that had been given a heavy focus Furthermore, another issue that

is current and prevalent specially among developing countries is the lower labour force participation (LFP) and high unemployment rates common among female population in these countries Depending on the availability of data, empirical-based studies offer important insights

on these issues and the results often vary because of sociological, household and economic characteristics that are prominent across these countries

Given the lacuna of Sri Lankan literature of the subject of interest, the literature survey conducted

is brief yet sheds some knowledge on the background of the labour market conditions in the country Also an emphasis is given to the empirical literature the models of paid and unpaid work offering a theoretical foundation which is useful in gaining an idea about the macro picture through micro-level analysis

The focus of the literature analysis aligns more towards women for two reasons This study is structured in a manner to identify the problem of lower female labour force participation and to identify if the latent factors that are dominated by sociological burdens Assisting this further, the literature review is laid as a pre-discussion to the analysis of results in later sections This literature review is organised in a thematic manner focusing on models of paid and unpaid work and on the empirical literature that looks into the LFP and unemployment issues in Sri Lanka

Models of paid and unpaid work

The models of time choices, paid and unpaid work is highly associated with time The evolution process has taken a leap since its theoretical foundation extracted from neo-classical economics restricting the use of time to merely utilitarian concepts However, this section does not attempt to deconstruct theoretical models of paid, unpaid and family models Alternatively, I attempt to identify various models hoping to provide an insight on how these models have been disrupted with dynamics of time, economy and sociological characteristics These models are briefly discussed in the literature review due to two important reasons; the first being to explore the economic rationale of a very broad, latent yet a serious issue in hindrance for the growth of women

as well as economies The second reason is to offer a basis to explore, suggest and the intensity of data restrictions that I face in structuring this study amidst of prevailing high dynamics of the society

Models of time Use in Paid and Unpaid Work by Beaujon & Liu (2005) and Mothers’ Time Choices – Caregiving, Leisure, Home Production and Paid Work by Kimmel & Connelly (2006) are significant among the observed literature due their in-depth analysis on the matter Both of these papers had analysed these models comprehensively and had adapted in a manner to fit their empirical needs and questions of interests of their studies

Beaujon & Liu (2005) in their study identifies three types of models for paid and unpaid work While identifying the significance of breadwinner models and double burden models on women, they had argued that the societies are evolving towards more novel ways taking into account the present positive trends such as shared roles between the two genders

According to prior studies conducted by the same authors (2000), they identify that the

“distribution of paid and unpaid” is in an evolutionary process because of changing roles that happens over an individual’s lifetime starting from role of the family, education level, age, attachment to the labour force and the cohort (Beaujot & Liu, 2005) Models with complementary roles could differ according to one highly emphasized characteristic Women or men who do more paid work, women or men who do less unpaid work, the double burden with separate categories for cases where women or men carry a larger amount of work, shared-role models where the unpaid work is more equally divided

Beaujot & Liu (2005) identifies three categories alternatively complementary roles based on dual-career options by families The conventional case where the husband is more inclined towards an ambitious career, but supportive with household and family care, gender-based specialization where the parenting is shared and the last model with role-sharing case with both equally involved

in career pursuits as well as household work

Complementing to both models, the paper further analyses the prevalence of quantitative models for the purpose of making empirical analysis convenient Such models are split models with 40/60 labour models which could be influenced through their share of earnings

All the above models including neo-classical model on time and choice are restricted to two dimensions of time allocations Alternatively, Kimmel & Connelly (2006), through expanding Gronau’s model improves the model further through introducing the wage effect, substitution effect and non-substitutable home production aspect This model could be a theoretical foundation

in understanding another dimension of female low LFP in Sri Lanka; the unaccounted investment they make for their child

As concluding remarks for this section, even though various models are structured in an attempt

to understand the dynamics in an ever changing environment, the objective of collating these models is to identify the macro picture with a theoretical basis when interpreting results As a result, it is evident that even though the bases are varied across these models, they are in fact complementary in identifying these dynamics

Literature on the issues of LFP and unemployment in Sri Lanka

Because of the lower number of empirical research on paid and unpaid work, this section attempts

to explore the nature of the labour supply and labour market constraints as to explore and

understand prevailing labour market issues in the country While a number of researchers in Sri Lanka have published several articles on unemployment, Rama Martin (1999) an economist from the World Bank was first out of the few who identified the importance of looking at the issue from

a different perspective; labour supply In his study, he identifies job search as one of the salient features in Sri Lankan labour market with proof A non-linear relationship between age and probability of being unemployed, selection of good jobs versus bad jobs, the “bimodal” relationship between income and unemployment where two peaks consisting of low levels of income and high income levels of income in the peak of unemployment provided needed evidence

in establishing his argument While plenty of studies identify the prevailing labour market rigidities, studies exploring the persistence of lower labour force participation among women is very limited

DeGraff & Malhotra (1997), one of the very descriptive papers on the issue of labour supply attempts to understand the behaviour of labour supply among young women through The study offers a comprehensive analogy of household, socio-economic characteristics that influences young women from entering the labour force in Sri Lanka The authors who focused on the labour force participation on young women, reveals important findings which could be explained through extracting historical facts Before the millennium, married women have had a higher labour force participation This can be justified through garment factories opening up new opportunities for low-skilled women in an era where economic turmoil often occurred The study finds that human capital, education and age are the most prominent factors in determining labour force participation among young women This finding aligns with theoretical explanations as well as indicating the supportive native of the family towards their daughter’s well-being Other than being a Moor discourages labour force participation due to their cultural perceptions, surprising results indicate that urban dwellers and households suffering from poverty have negative influences on labour force participation among young women This could be due to poor networks of young females or perhaps due to added weight on the young daughter to take care of their younger siblings in economic conditions where mother’s non-skilled labour is significantly recognized in the labour market

Gunatilake (2013) in her report identifies trends and patterns of labour force participation among females in the country Her approach to address the data restrictions had been to identify different factors through representative sample survey highlighting the contribution for the empirical literature by conducting the first representative survey centred around labour force supply among women Gunatilake (2013) finds out that remittances from abroad, Moor ethnicity, women with higher household earnings and marriage had restricted women from entering the female labour force participation The study further concludes that 68% of the married women’s participation are indicative through demographic and educational characteristics while 10% of the labour force participation constitutes through household characteristics

By analysing the brief survey of literature, it is evident that the labour economists have attempted

to address this contemporary issue using two approaches The first approach is through identifying various family models through empirical data and the other approach taken is identifying various determinants of paid and unpaid work Analysis of the prior methodology requires very

comprehensive data and offers a glimpse of the social structures prevalent which can be very useful

in structuring very sound labour policies This could be the first approach in

1 Identifying if there exist labour market rigidities which discourages women from entering the labour force

2 If it’s the social perceptions and norms that restricts women from entering the labour force

Research Methodology

Conceptualisation

1 Identification of individuals who are not in the labour force

The population of the country consists of two segments; ones who are in the labour force and ones who are not in the labour force Labour force in Sri Lanka is identified as the population economically active consisting of 15 years of age and above (Refer diagram 1) An individual is identified as not part of the labour force as follows;

1) Determining if the individual is economically active (LFS question no 2 and question no.4) 2) Determining if the individual is presently seeking employment options (LFS question no 42)

3) If the individual is not seeking employment options, he or she is categorised as an individual not in the labour force

2 Paid work and unpaid work

Unpaid work is an area that is rarely explored in the labour market research Therefore, in this study, unpaid work is very narrowly defined as persons who are not in the labour force due to the reason of household work The hours spent on household work, sub-categories are important sections but are not recorded in the present LFS

Diagram 1 – Conceptualising and identifying the population sample

Is the individual currently employed?

• NO

Is the individual presently looking for employment options?

• NO

What is the reason for not seeking employment?

•Household work

Theoretical framework

Despite various models of paid and unpaid work identified in the literature review, the theoretical framework for this study is based on the theory of neo-classical labour supply which takes a utilitarian approach in determining the labour supply

Given the 24-hour time constraint, an individual maximises their utility through choosing number

of hours allocated for work and leisure (Equation 1)

𝑡𝑖= 𝛽1𝑤𝑜𝑟𝑘𝑖+ 𝛽2𝑙𝑒𝑖𝑠𝑢𝑟𝑒𝑖 (1)

where 𝑡𝑖= time an individual has per day, 𝛽1𝑤𝑜𝑟𝑘𝑖 and 𝛽2𝑙𝑒𝑖𝑠𝑢𝑟𝑒𝑖 indicates the number of hours allocated for each activity

The decision of allocating the number of hours per each activity depends on the utility an individual gain through engaging in that particular activity Under the assumption that hourly wages or the expected hourly reservation wage is an accurate measurement for the utility of an individual, he or she would enter the labour market if their utility of the hourly wage (reservation wage or earned wages) exceeds the utility of their hourly leisure time Therefore, we can estimate that an individual would make the decision to enter the labour market depending on equation (2) and (3)

𝑙𝑓𝑝𝑟 = 1 𝑖𝑓 ℎ𝑜𝑢𝑟𝑙𝑦 𝑤𝑎𝑔𝑒 > 𝑢𝑡𝑖𝑙𝑖𝑡𝑦 𝑜𝑓 𝑙𝑒𝑖𝑠𝑢𝑟𝑒 (2) 𝑙𝑓𝑝𝑟 = 0 𝑖𝑓 ℎ𝑜𝑢𝑟𝑙𝑦 𝑤𝑎𝑔𝑒 < 𝑢𝑡𝑖𝑙𝑖𝑡𝑦 𝑜𝑓 𝑙𝑒𝑖𝑠𝑢𝑟𝑒 (3) where 𝑙𝑓𝑝𝑟 indicates the individual’s decision on entering the labour force If 𝑙𝑓𝑝𝑟 = 1, an individual seek employment given that their utility of hourly wage is higher than their utility of leisure 𝑙𝑓𝑝𝑟 = 0 indicates that the individual decides not to enter the labour force given that their utility of leisure is higher than their given or expected wage

Due to the finite researches available on this area of research, this study offers a prerequisite for further exploration of gendered segregation Given this stance, I hope to identify this identified prevalence of lower labour force participation among women is caused because of the

predetermined role of a mother and a wife

Estimation and the econometric model

Previous studies are structured in a manner that facilitates the availability of the microdata relevant

to the study The literature examined within the scope of this study uses continuous dependent variables such as hourly wages in most studies However, because of data constraints even within the range of microdata restricts this study from examining other factors which are further discussed

in the section on limitations

Due to the area of interest centred upon labour force participation, unemployment, paid and unpaid work, the model used to answer the research questions is the bivariate probit model

Equation

1) Univariate probit regression

As the initial step is to identify if there is a significant gendered segregation in deciding paid and unpaid work in Sri Lanka Therefore, a univariate probit model is estimated using limited dependent variable unpaid work, a vector of individual, household and sociological characteristics are used as dependent and independent variables respectively As equation (4) indicates,

𝑝𝑟𝑜𝑏(𝑢𝑛𝑝𝑎𝑖𝑑𝑤𝑜𝑟𝑘 = 1) =∝ +𝛽𝐹𝑒𝑚𝑎𝑙𝑒𝑠 + 𝛽2𝑋2+ 𝑢𝑖1 (4)

the probability of unpaid work vs paid work is expected to be estimated through ∝ - the constant, 𝐹𝑒𝑚𝑎𝑙𝑒- the dummy variable for females, 𝑋2 representing a vector of other variables and 𝑢𝑖 indicating the error term

2) Bivariate probit regression

Even though equation (4) could be used to identify determinants of unpaid work, measurement errors can incur because of the selectivity biasedness Selectivity biasedness occurs when the impact of independent variables on the dependent variable is estimated for a particular subset of the population This would result in inaccurate marginal effects from the regression analysis

In order to address this issue, bivariate probit regression is used In addition to addressing the selectivity biasedness, estimating the bivariate probit regression will also be capable of identifying the impact of unpaid work on LFP within the labour market in Sri Lanka Bivariate probit model

is estimated in two steps;

Equation 1 𝑝𝑟𝑜𝑏(𝑢𝑛𝑝𝑎𝑖𝑑𝑤𝑜𝑟𝑘 = 1) =∝ +𝛽𝐹𝑒𝑚𝑎𝑙𝑒𝑠 + 𝛽2𝑋2+ 𝑢𝑖 (4)2

where 𝑝𝑟𝑜𝑏(𝑛𝑜𝑡𝑤𝑟𝑘𝑓𝑟𝑐𝑒) = 1 𝑖𝑓 𝑝𝑟𝑜𝑏(𝑢𝑛𝑝𝑎𝑖𝑑𝑤𝑜𝑟𝑘) = 1

𝑝𝑟𝑜𝑏(𝑛𝑜𝑡𝑤𝑟𝑘𝑓𝑟𝑐𝑒) = 0 𝑖𝑓 𝑝𝑟𝑜𝑏(𝑢𝑛𝑝𝑎𝑖𝑑𝑤𝑜𝑟𝑘) = 0 Equation 2 𝑝𝑟𝑜𝑏(𝑛𝑜𝑡_𝑤𝑟𝑘𝑓𝑟𝑐𝑒 = 1) =∝ +𝛽1𝑢𝑛𝑝𝑎𝑖𝑑𝑤𝑜𝑟𝑘 + 𝛽2𝑋2+ 𝑢𝑖 (5)3

In an attempt to understand the impact of unpaid work on not entering the labour force

participation, following is estimated used biprobit command in STATA statistical analysis

software

Use of interaction effects

In the case of identifying sociological aspects, the assumption ceteris paribus overlooks combined

effects of certain sociological characteristics For example, a Moor female might be more pressured or inclined towards choosing household care over labour force participation To further explore the intensity of these sociological traits, interaction effects are used to identify the

1 Model adapted from Gender Segregation in Education and Its Implications for Labour Market Outcomes: Evidence from India to suit the needs of the study

2 Model adapted from Chisadza (2015) accordingly

3 Model adapted from Chisadza (2015) accordingly

combined effects of the most significant in the previously estimated equation (5) Interaction effects (equation 6) are estimated as follows;

𝑝𝑟𝑜𝑏(𝑢𝑛𝑝𝑎𝑖𝑑𝑤𝑜𝑟𝑘 = 1) =∝ +𝛽𝐹𝑒𝑚𝑎𝑙𝑒𝑠 + 𝛽2𝑋2+ 𝛽3𝐹𝑒𝑚𝑎𝑙𝑒𝑠 𝑥𝑖+ 𝑢𝑖 (6)

In the equation (6), the interaction effect is identified through variable 𝛽3𝐹𝑒𝑚𝑎𝑙𝑒𝑠 𝑥2𝑖 where 𝑥2𝑖is

a dominant variable of vector 𝑋2 of equation (5)

Limitations

Limitations of this study is mainly caused through insufficient availability of data The following are some of the limitations identified within the scope of this study

1 Unpaid work is restricted only to household care Therefore, this variable cannot be analysed in depth Hence, one of the reasons this study restricted itself through a probit model approach This is one of the major flaws that needs to be addressed in the next labour force surveys

2 Reservation wages in the labour force survey are only for those are seeking employment Therefore, due to another data limitation, the impact of wages is not captured at all

3 Since the literature on this subject of interest is scarce, the emphasis was made only to identify if there exists a gendered segregation which discourages one from labour force participation Therefore, the exclusion of the variable of earnings of the partner is another limitation in the study

4 The variable dependency could be further broken down according to age, type of the dependent; child or parent Due to time limitations, this was one aspect which was not taken into account

Sources of data

The only source of data is the microdata retrieved from Labour Force Survey (LFS) conducted by the Department of Census and Statistics (DCS) in 2016 This is the most updated set of microdata which is readily available

Dependent variable

- Paid work 1= Individuals who are presently in the labour force (employed &

unemployed

- Unpaid work 1= Individuals who are presently not in the labour force

Independent variables

- Female Dummy variable (Females=1) Reference category “males”

- Marital status 4 categories (married, widowed, divorced & separated) Reference

category “single”

- Ethnicity A categorical variable with 7 categories (Sri Lankan Tamil, Indian

Tamil, Sri Lankan Moor, Malay, Burger, Other) Reference category

“Sinhalese”