THESIS PROPOSAL
NAME, ID
Title: Family-friendly work policies and women's labour supply following maternity
Supervisors: A/Prof. E. Magnani, Prof. D. Fiebig
Other Academics with whom I wish to discuss my topic: A/Prof. G. Barrett
General Area: I intend to investigate the area of Labour Economics, particularly factors influencing women’s labour supply decisions after having children. I am particularly interested in the effect that family-friendly work policies have on decisions to return to the labour force, job mobility and consequent impacts upon women’s skills.
Data: The main sources of Australian data are:
1. Household Income and Labour Dynamics Survey (HILDA)- this is a panel data set which began with 7982 households and 19914 individuals in Wave 1, and has just released Wave 7.
· Waves 1-5 include 218 observations of a birth/adoption event which has employment observations for up to 2 years before the event, and employment observations for up to 3 years after the event (Hosking, 2007). On an averaged basis, approximately 262 birth event transitions are expected to be obtainable from Waves 1-6. Waves 1-5 also contains 6723 non birth-related work transitions for 1840 women (Hosking, 2007). Thus, 8068 observations are expected from Waves 1-6. This will enable a comparison of factors influencing mobility which differentially impact upon women without children, and women with children of various ages.
· It includes unit file information on employment history, including work hours transitions and job transitions; information relating to fertility, personal history, relationships, and household / family circumstances. These variables will enable a rounded assessment of factors influencing labour supply decisions (particularly with information about partners’ employment and the availability of services such as childcare).
· As HILDA has information about women of all ages and circumstances, it enables:
a) analysis of the endogeneity of the fertility decision, and
b) analysis of women’s self-selection into employment with flexible work opportunities.
2. Parental Leave in Australia Survey – this is a supplement to the 2005 Longitudinal Survey of Australian Children (LSAC).
· It includes 3568 unit file responses for employment; the use of leave by mothers and partners; and working arrangements for both parents both before, during, and after the pregnancy.
· As the survey was only distributed to the infant cohort, there is a selection bias as only parents with newborn are interviewed.
3. Pregnancy and Employment Transitions Survey – this 2005 ABS dataset is a cross-sectional survey of women with children under 2 years of age.
· It includes information about employment status and employer information before, during, and after pregnancy, including information on the use of leave, childcare arrangements, and reasons for labour supply decisions following birth.
· Information is not unit-file, therefore there cannot be an analysis of individual characteristics’ influence on labour supply decisions.
Literature: The key literature can be categorised into the following relevant areas:
1. Labour Mobility and Leave Arrangements:
· Whitehouse, G. & Hosking, A. (2005) “Policy Framework and Parental Employment: A comparison of Australia, the United States, and the United Kingdom” http://melbourneinstitute.com/hilda/Biblio/cp/gwah_05_fin.pdf, accessed 20 March 2008
o Provides descriptive statistics from the Parental Leave in Australia Survey, as well as a review of the literature.
· Hosking, A. (2007), “The Effects of Motherhood and Job Transitions on Female Earnings in Australia,” Conference paper for the non-refereed stream of the 2007 HILDA Survey Research Conference, 19-20 July, The University of Melbourne.
o Methodological information on issues specific to HILDA (particularly combining the waves of data, and isolating variables indicating pregnancy and maternity leave). Hosking employs Mincerian regression models (first-differenced fixed effects) to test for the impact of job transitions on wages. For birth-related job transitions, she isolates observations with employment data in the periods before and after the birth year, which is an approach I am considering.
· McRae, S. (1994) “Labour Supply After Childbirth: Do Employers’ Policies Make a Difference?” Sociology 28(1) 99-122
o McRae employs matched employer-employee data to analyse the influence of work and family characteristics on three variables: the probability of returning to work within 9 months of birth (logit); hours of work following birth (ordered probit); and the probability of returning to the pre-birth employer (bivariate probit). She finds that family friendly work policies, with the exception of jobsharing, have an economically and statistically insignificant effect on the hours of work and the probability of returning to the pre-birth employer. However, McRae does not test for job-specific factors influencing the availability of leave which may influence women’s initial choice of employer, which may influence the effects of the employer variables.
· Rhum, C.J. (1990) “Bridge Jobs and Partial Retirement,” Journal of Labour Economics, Vol. 8 No. 4 (Oct., 1990) pp 482-501
o Rhum employs panel data to analyse the effect of various personal and work-related characteristic (including the availability of pension policies) on the labour transitions of older workers from full-time careers jobs to full retirement. The three analyses are: the age at which people leave their career job (ols); whether a bridge job is undertaken (logit); and the years of post-career labour force participation (Cox proportional hazards). He finds that the majority of workers switch to ‘bridge jobs’ in the transition phase, with most workers switching to a different industry or occupation. Aspects of the transition, however, depend upon industry, skills and employer policies.
2. Fertility:
· Weston et al (2005) It’s not for a lack of wanting children: A Report on the Fertility Decision Making Project, Australian Institute of Family Studies
o Highlights the factors influencing parents in their decision to have a child.
o Provides a reference for the variables required in an analysis of the pregnancy decision.
o Provides rankings of each influence upon fertility, as determined by men and women. Also provides further demographic breakdowns of preferences.
· Risse, L. (undated) “Determinants of Maternity Leave Provisions In Australia and the Effects on Fertility: An Application of the Heckprobit Selection Model,” Paper Submission for Peer Review, ACSPRI Social Science Methodology Conference HILDA Stream
o Risse uses HILDA Wave 3 to analyse: the probability of maternity leave provision (probit model with a Heckman selection variable addressing the probability of knowing about leave provisions); and the effects of maternity leave on fertility (probit). She finds that “paid ML is more likely to be provided to women in permanent employment, the public sector, the unionised workforce, larger firms and highly-skilled occupations.” (p19) Her analysis of the impact of employer policies on fertility is very useful for my research, as it will help to address endogeneity in assessing whether factors influencing the return to work also influence the fertility decision.
Specific Topic and Motivations
The crux of my interest is to investigate the ways in which women’s labour force participation and human capital can be maintained after childbirth. I have chosen to address this through an investigation of whether flexible work policies offered by employers encourage this. While the Australian debate surrounding maternity leave is focussed on encouraging fertility, I believe there has not been enough focus on how women can be assisted in the longer term, which restricts women’s choices regarding work and family, and risks the loss of women’s human capital.
In an era of skills shortages and declining fertility, Australia needs to minimise the barriers to work and encourage a more flexible workplace culture. This was recently highlighted by the Chamber of Commerce and Industry, which suggests that flexible working hours are the key to encouraging women into the workforce (ABC News Online, Tue 25th March 2008). While this has been assisted by moves to increase the casualisation of the workforce, the associated lack of job security and benefits is likely to be detrimental to fertility decisions, which are heavily influenced by the ability to afford a child (AIFS 2004).
Issues Regarding the Data
Background reading suggests that family-friendly work policies are usually offered by larger organisations attempting to retain skilled workers. This employer-led approach limits such options to women who, by nature of their higher human capital accumulation (through education and labour force experience) are in any case more likely to re-enter the workforce. Thus there is a risk that there are insufficient observations of lower-skilled occupations to facilitate an appropriate analysis of employer policies after controlling for skills and education (see McRae, 1994).
In this situation, I may need to separately analyse the patterns of return to work depending on factors which characterise groups of women who have differential policy access. This may be approach through the use of instrumental variables which address the probability of family-friendly polices being offered. Overall, I intend to analyse factors influencing women’s labour force participation decisions in three periods: before, during, and after pregnancy. This information will be collected from the various periods and pooled to create cross-section for analysis. I intend to employ logit and/or probit models to analyse the influence of the relevant variables, following the approach of Rhum (1990), McRae (1994) and Risse (undated).