Summary

This paper presents new Australian research on how different factors relate to the timing of women's return to work after having a child, contributing significantly to our understanding of maternal employment in Australia. The focus is on how different aspects of employment prior to a birth, and the use of paid and unpaid leave, are related to the timing of return to work after a birth. By exploring relationships between employment characteristics, use of leave and timing of return to work, this paper contributes to the current policy debate on paid maternity leave (Goward, 2002; Productivity Commission, 2008; Swan, Gillard, & Macklin, 2008) and also informs on how labour market features impact on this aspect of family life.

The analysis is based on data from the 2005 Parental Leave in Australia Survey (PLAS), which was conducted as a nested survey in Wave 1.5 of the Longitudinal Survey of Australian Children (LSAC). This national survey, which contained a large range of pre-birth and post-birth employment details, had a sample size of 3,573 respondents, providing an excellent opportunity to analyse contemporary patterns of maternal return to work. The sample, however, slightly over-represented more highly educated women with a greater connection to the labour market. The results therefore do not necessarily reflect the employment behaviour of all Australian women.

The timing of mothers' return to work following childbearing is related to a wide range of variables, including mothers' preferences and opinions about the appropriateness of their remaining home when children are young. Other work-related factors can also be important in explaining variations in the timing of return to work. For example, some mothers may be able and may wish to return to work very soon after the birth of a child because they have access to a job with flexible work conditions or very short hours that can fit around responsibilities for caring for an infant. Other mothers may have access to sufficient leave to enable them to take an extended break from employment through the child's first year. For others, managing work during the months following a birth may not be so easy. If paid leave is either not available or available for an insufficient time, some mothers may return to work sooner than desired. Others may not have access to leave, having worked in sporadic casual employment prior to the birth. Still others may have difficulties in finding a job that offers the flexibility that is required in order to manage work and family.

Differences in the timing of return to work are expected according to a range of parental characteristics. The main focus of the paper is on the extent to which pre-birth job characteristics (in particular, type of employment contract) and use of paid and unpaid leave are associated with differences in the timing of return to work within the first 18 months after the birth of a child. While the analysis presented here does not extend to the types of jobs mothers return to, or their reasons for returning to work (or remaining out of work), the results contribute to our understanding of which women experience a relatively quick return to work, and which women delay their return to work more than others. From a policy perspective, these groups of women are of interest. Among the former, there may be some women (and their children) who are at increased risk of experiencing more negative health and wellbeing outcomes (Berger, Hill, & Waldfogel, 2005; Brooks-Gunn, Han, & Waldfogel, 2002; Hyde, Klein, Essex, & Clark, 1995). Among the latter, there may be mothers who will find it more difficult to re-enter the labour market as their child gets older, which is likely to affect their longer-term financial wellbeing (Arun, Arun, & Borooah, 2004; Beggs & Chapman, 1988; Breusch & Gray, 2004).

For this paper, the timing of return to work was first examined by calculating the percentage of women who had returned to work in each month up until 18 months after the birth of the child. Only 2% of mothers were in paid employment with a child aged less than one month, 11% had returned to work by the time their child was three months old, and 22% had returned when their child was aged six months old. By the time the child was 12 months old, 44% had returned and by 18 months, 54% had returned.

The likelihood of mothers returning to work in a given month was also analysed in this paper. This measure, typically referred to as the "hazard" of returning to work, is a useful measure of the peak times of returning to work. From child ages of 2 to 8 months, of those who had not yet returned to work, around 4% returned each month. There was a slight increase in the likelihood of returning at 9 to 10 months, and then a peak of around 7% when the child was aged 11 months. The likelihood of returning remained high for children aged 12 months (6%) but then fell in subsequent months. Returning to work when the child was aged 11 to 12 months is likely to be associated with the availability of one year's unpaid maternity leave for women who were in permanent employment before the birth.

Two-thirds of women were employed in the 12 months prior to the birth of the study child. These women were most commonly working as permanent employees (43% of all women). Casual employees, contract employees and self-employed women made up 14%, 3% and 7% of the women respectively. Being employed prior to the birth was a very strong predictor of the timing of return to employment after the child's birth. Those who had not been employed during their pregnancy experienced the lowest rate of return to work. Just 13% in this group had returned by 18 months after the birth. Among those who were employed prior to the birth, the self-employed were the most likely to have returned 18 months after the birth (84%), followed by permanent employees (76%), contract workers (64%) and casual employees (58%). At 12 months, while fewer had returned, the differences by employment type were still evident. At 9 months and earlier, however, there were far smaller differences between those who were previously permanent, casual or contract employees. The self-employed, albeit a small proportion (7%) of the women, were more likely to return to work than others, especially while the infant was very young. They had a very high likelihood of returning to work when the child was aged less than 3 months (53% returned).

Around three-quarters of women employed during pregnancy used some leave after the birth of the child. Just 10% used paid leave only, but 37% used paid and unpaid leave and another 25% used unpaid leave only. The majority of permanent employees used some leave after the birth, with just 12% using no leave. Most permanent employees used some paid leave (69%) and another 19% used unpaid leave only. Casual, contract and self-employed workers were much more likely to have used no leave (about 56%) and more likely, if they did use leave, to have used unpaid leave.

Using leave was related to higher rates of return to employment, although there were different effects over the first 12 months. In particular, in the first three months, those who used a combination of paid and unpaid leave were less likely to return to work than other leave-users, no doubt because this combination of leave enabled them to take a longer absence. These women had the highest rate of return to work between months 10 and 12. From this time, mothers who used a combination of paid and unpaid leave did not have significantly different rates of return to work to those who used only paid leave. Across all months, those who used only paid leave tended to have a higher rate of return to work than those who used only unpaid leave, although by 18 months, the differences were quite small. Taking only paid leave rather than only unpaid leave was associated with a slightly higher rate of return to work (83% and 77% respectively had returned by 18 months after the birth). Of those who used a combination of paid and unpaid leave, 80% had returned by 18 months after the birth.

The higher rate of return to work of permanently employed women appears to be related to their use of leave, since these women were much more likely than others to have used leave, but in particular, to have used paid leave or a combination of paid and unpaid leave. For example, the peak time of return to work among women who were previously permanent employees was at 11 to 12 months, which would equate to when a period of leave might end.

This analysis provides more detailed insights into maternal employment following childbirth than is possible from conventional analyses of cross-sectional data. By determining which groups of women have a slower return to work - for example, those who were not employed during pregnancy or who were employed but took no leave - we can identify who might face longer-term difficulties in the labour market. Also, by identifying those who return to work early - for example, those who return to self-employment - we can consider whether some of these women might face particular maternal or child health issues because they return to work soon after the birth. Some of these issues require more thorough examination, taking into account women's reasons for returning to or remaining out of work, and the characteristics of their jobs on their return. These details are available in the PLAS and will be the subject of future analyses of these data. Further, the comprehensive information on child and maternal characteristics, along with the longitudinal nature of the LSAC data, will allow follow-up of how these post-birth employment patterns are associated with outcomes for children and mothers.

Introduction

Despite increasing rates of maternal employment in Australia, there is little Australian research on the employment transitions of mothers in the months following the birth of a child. This paper addresses this by providing an overview of the timing of maternal return to work, and exploring how this timing varies with different pre-birth employment characteristics and types of leave.

Australia is one of a few OECD countries without universal entitlement to paid maternity leave (Organisation for Economic Co-operation and Development, 2007), and the diversity of the labour market means that some women are employed in jobs that provide access to paid and unpaid maternity leave, while others in more tenuous positions receive only unpaid maternity leave or perhaps no leave at all. This is in contrast to what the OECD report labels a "good leave scheme", which gives parents choice and flexibility in their return-to-work decisions (OECD, 2007, p. 21). This may be one reason for the employment rate of mothers with young children being lower in Australia than it is in many other OECD countries (OECD, 2007, p. 16).

By exploring relationships between employment characteristics, use of leave and timing of return to work, this paper contributes to the current policy debate on paid maternity leave (Goward, 2002; Swan, Gillard, & Macklin, 2008) and also informs on how labour market features impact on this aspect of family life. It is important to understand these relationships, as the timing of mothers' return to work is likely to have implications for their longer-term connection to the labour market and finances (Arun, Arun, & Borooah, 2004; Beggs & Chapman, 1988; Breusch & Gray, 2004) and also has relevance to issues of child and maternal health and wellbeing (Berger, Hill, & Waldfogel, 2005; Brooks-Gunn, Han, & Waldfogel, 2002; Hyde, Klein, Essex, & Clark, 1995).

Analyses of maternal employment transitions have been presented for Australia, but these were based on older (1985) data (Glezer, 1988) or on annual, rather than monthly, employment information (Baxter, 2005). Glezer's and Baxter's analyses showed that women employed during pregnancy returned to work sooner after childbirth than women who were not employed during pregnancy. Further, Glezer's study showed that women who used maternity leave had a higher retention rate in employment. This was not surprising, since those who did not use maternity leave included those who left work with the intention of taking time out of the labour market to care for their children full-time. This will include those who did not have a permanent job before the birth and did not have access to paid or unpaid maternity leave. It is clear that this finding should be reassessed, as these data are now more than twenty years old. Also, while different job conditions during pregnancy were studied in Glezer's analysis, little is known about whether, in the current labour market, they lead to differences in return-to-work patterns. International research can be informative but may not necessarily apply, given that parental leave entitlements and employment arrangements in Australia differ to those of other countries.

Data collected recently have made it possible to undertake analyses of contemporary return-to-work timing, taking into account the use of leave as well as other factors such as job characteristics prior to the birth. This analysis uses data from the 2005 Parental Leave in Australia Survey (PLAS) (Whitehouse, Baird, & Diamond, 2005), a nested study within Wave 1.5 of the Longitudinal Study of Australian Children (LSAC). Using these data, this paper provides new evidence of how different factors relate to the timing of women's return to work after having a child, contributing significantly to our understanding of maternal employment in Australia. This paper considers the extent to which pre-birth job characteristics (in particular, type of employment contract) and paid and unpaid leave (maternity or other) are associated with differences in the timing of return to work within the first 18 months after the birth of a child.

After a review of literature and an overview of the data and methods used, the results are presented. First, an overview of pre-birth employment and leave use is presented as a background to the analyses of return-to-work timing. This is followed by a description of the return-to-work patterns of all mothers. The paper's next two sections explore how return-to-work timing varies with type of employment contract during pregnancy (referred to throughout as pre-birth employment type) and use of leave. A further section then summarises the associations that were found for the other demographic and employment variables included in these analyses. The paper finishes with a discussion of the findings and a conclusion.

Literature

When making decisions about returning to work, women are likely to consider a multitude of factors, not least being the need or wish to remain at home with children and the availability of alternate child care. Employment-related factors, including the opportunity cost of remaining at home with children, the immediate need for the contribution of the women's income, and the extent to which women feel they need to maintain a connection to the labour market or a particular job or career are likely to be important.

Factors associated with maternal employment transitions have been explored for a number of countries, and presented in the international literature. The following review examines some of these papers. United Kingdom data were used in analyses by Macran, Joshi, and Dex (1996) and Dex, Joshi, Macran, and McCulloch (1998). Hofferth (1996), Joesch (1994, 1997), Lyness, Thompson, Francesco, and Judiesch (1999) and McGovern et al. (2000) used data from the United States. Ondrich, Spiess, Yang, and Wagner (1999) focused on Germany. Comparative studies include those of Rønsen and Sundström (2002) for data on Finland, Norway and Sweden; Pylkkänen and Smith (2003) for data on Denmark and Sweden; and Waldfogel, Higuchi, and Abe (1999) for data on the United States, Britain and Japan.

Studies consistently have found that having worked before the birth of a child is strongly associated with working after the birth, and returning to work faster (Baxter, 2005; Baxter, Gray, Alexander, Strazdins, & Bittman, 2007; Glezer, 1988; Hofferth, 1996; Joesch, 1994).

For those employed prior to having a child, the availability of parental leave is relevant to the process of decision-making about the timing of return to work. Parental leave assists in maintaining a connection to a job and, when paid, addresses financial needs. Access to leave, whether paid or unpaid, is expected to have a positive effect on the likelihood of returning to work after a birth, since it guarantees a job at the end of the period of leave (Hofferth, 1996; McGovern et al., 2000; Pylkkänen & Smith, 2003; Waldfogel et al., 1999). A longer period of leave, however, may mean women take longer to return to work (Rønsen & Sundström, 2002) and a longer absence from the workplace reduces the likelihood of return at the end of that leave period (Ondrich et al., 1999). Whether leave is paid or unpaid may make a difference to return-to-work timing, since a period of paid leave provides more of an incentive to remain at home than a period of unpaid leave. For example, McGovern et al. (2000) found that a longer period of paid leave was associated with a longer break from work. Joesch (1997), on the other hand, found that paid leave was linked to a faster return to work once the child was at least two months old.

In Australia, one year's unpaid maternity leave is available to permanent employees with one year's continuous employment with an employer prior to the birth. Casual workers' entitlements to unpaid maternity leave depend on a variety of factors, including the length of time with their employer (Whitehouse, Baird, & Hosking, 2007). Access to paid maternity leave is not universal and availability and length of leave differs from one job to the next as a condition of employment. Workers may also use other paid leave, such as holiday leave, around the birth of a child. Paid leave is unlikely to be offered to casual employees, who usually have no paid leave entitlements. Those who are not employed during pregnancy have no entitlement to maternity leave, paid or unpaid. For further discussion of paid and unpaid maternity leave in Australia, also refer to Edwards (2006).

According to the recent Pregnancy and Employment Transitions Survey (PaETS) in Australia (Australian Bureau of Statistics [ABS], 2006), 74% of the women with a two-year-old child who had a job while pregnant took some leave when they had the child (13% took paid leave only, 27% unpaid leave only and 34% both paid and unpaid leave). Using the Parental Leave in Australia Survey, Whitehouse et al. (2007) found that, of mothers who were continuously employed with the same employer for the 12 months before having the child, 46% used paid maternity leave (for an average length of 11 weeks), 68% used unpaid maternity leave (for an average length of 35 weeks), 46% used other paid leave and 12% used other unpaid leave.

Return-to-work patterns are likely to vary according to the conditions of the pre-birth employment. For example, greater access to family-friendly work arrangements may be associated with a faster return to work. Since hours worked, occupation, sector and size of business have associations with the amount of work-family strain (Alexander & Baxter, 2005), these characteristics might also be associated with return-to-work timing. These and other characteristics may also capture possible differences in job commitment (Lyness et al., 1999; McGovern et al., 2000). For example, size of business may be important, in that women employed in small businesses may feel a stronger attachment to work; however, women employed in larger businesses may have greater access to leave and other job conditions that ease work-family conflict. Also, the Pregnancy and Employment Transitions Survey showed that the types and amounts of leave used varied across different types of employment. For example, use of paid leave and total leave duration was greater for women who had worked in the public sector, were in higher-status occupations, had longer job tenure and worked in businesses with more than 20 employees (ABS, 2007). This suggests that job characteristics could be an important factor in explaining variations in return-to-work timing.

Various other factors are likely to be associated with differences in patterns of return to work. A higher level of education is usually associated with a faster return to work (Baxter, 2005; Dex et al., 1998; Hofferth, 1996; Macran et al., 1996; Polachek & Sieber, 1993; Pylkkänen & Smith, 2003). To some extent this reflects differences in preferences for employment, and also the opportunity costs of not working, which are likely to be higher among those with a higher education. Mothers' education level may also capture variations in opportunities to return to work, or in constraints due to limited access to jobs or child care. Mothers' return-to-work timing is also expected to vary according to relationship status and partners' employment status and income level, as well as family size and region of residence (Gray, Qu, de Vaus, & Millward, 2002; Hofferth, 1996; Joesch, 1994; McGovern et al., 2000; Miller, 1993).

Data and methods

The Parental Leave in Australia Survey

Conducted in 2005, the PLAS (Whitehouse et al., 2005) was nested within the Wave 1.5 collection of the LSAC.¹ The PLAS was directed only to parents of the LSAC's infant cohort and collected information on employment before and after the birth of the "study child" - the child that was the focus of the initial LSAC study. The aim of the PLAS was to fill gaps in knowledge about the use of parental leave in Australia. It collected information about parents' work history and employment characteristics around having children, including return-to-work information. Given this extensive range of information and a large sample size of 3,573 respondents, this survey provided an excellent opportunity to analyse contemporary patterns of maternal return to work. The children, at the time of the PLAS, were aged between 15 and 29 months. The majority of children were aged at least 18 months, so analyses of the timing of return to work within 18 months after the birth was possible.

One issue with the PLAS is that it is not representative of all mothers in Australia. The Wave 1 LSAC sample, from which the PLAS sample was drawn, was selected to be broadly representative of the Australian population, but due to non-response was slightly biased towards more highly educated women (Australian Institute of Family Studies, 2005; Soloff et al., 2005). The PLAS response rate was 71% (as a percentage of questionnaires sent), and introduced further bias towards more highly educated women. For example, according to the population census in 2001, 57% of mothers with a child under the age of one had completed year 12 schooling,² but among the LSAC mothers of the infant cohort, 67% had completed year 12, and in the PLAS, 73% had completed year 12.

Further differences between Wave 1 of the LSAC and the PLAS were evident according to mothers' employment status. Response rates were higher for the PLAS among women who, at Wave 1, were on leave from a job (81% responded) or employed and back at work (73% responded). The response rates were lower for women who were not in the labour force (63%) or unemployed (45%) at Wave 1. Furthermore, using the Wave 1 data on whether women were employed during pregnancy revealed that response rates to the PLAS were higher for those employed during pregnancy (74%) than for those who were not (59%). This analysis, therefore, is based on a dataset that under-represents those with less connection to the labour force, both before and after the birth, and as a result overstates the proportion returning to work relative to the entire population of women having children.

The return-to-work measures

The key data item in this analysis was the age of the child at mother's first return to work after the birth. This measure was derived from information provided by respondents when asked for the month and year they returned to work (if they had returned) and comparing this date to the child's birth date. Respondents were excluded from the analysis if the age of the child on their mother's first return to work was unable to be determined (102 respondents). Women who had not returned to work by the time of the survey were included in all analyses relating to return to work up to the age of their child, but excluded for other analyses. For example, cases were excluded from analyses of returns between child ages of 13 and 18 months if the child was not yet 18 months old at the time of the survey and the mother had not yet returned to work (because she may or may not have returned to work by the time the child reached 18 months old).

The timing of return to work was first examined by calculating the percentage of women who had returned to work ("returned" is used even though some may never have worked) in each month up until 18 months after the birth of the child. Secondly, the "hazard" of return (the likelihood or risk of returning to work in a month, given not previously having returned to work) was analysed. It is a useful measure of the peak times of return to work.

To analyse correlates of return to work, multivariate analyses were used to isolate relationships with pre-birth employment status and use of leave, while controlling for other factors (see the next section for more details). Pre-birth employment status classified women as being not employed, employed in a permanent, casual or contract (fixed-term) job, or self-employed before the birth of the study child. This was based on women's own reporting of which category best described their employment situation, when asked about the 12 months before the study child was born. Those not reporting to be self-employed were given the options of "permanent employee (with an expectation of ongoing employment and access to paid annual and sick leave)", "casual employee (employed on a casual basis, without access to paid annual and sick leave)" and "fixed-term employee (on a contract with a set completion date)".³

Mothers' use of leave was classified as "unpaid leave only", "paid leave only", "both paid and unpaid leave", and "no leave". Unpaid leave included unpaid maternity as well as unpaid other leave, and similarly paid leave included paid maternity as well as other paid leave. Some respondents were excluded from the analysis if they reported that they used leave but provided no details of the types of leave used (that is, whether it was paid or unpaid). The "no leave" category is quite heterogeneous, including those who took no leave but returned to work straight away, as well as those who left their job. Of all women who were employed during pregnancy but took no leave, half said they took no leave because they left (49%) or were dismissed from (6%) their job while they were pregnant. Another 17% said it was because they were self-employed, 9% said they had no access to leave, 6% said their job was flexible enough to not need leave, and 15% cited other reasons.

Use of leave rather than entitlement to leave was included in this analysis, as the survey did not collect information on leave entitlements. Analysing return-to-work timing according to leave use is somewhat problematic, as decisions about types of leave used are woven into decisions about the length of time away from employment. For example, women who used only paid leave are likely to have had a shorter absence from work than those who used some unpaid leave, as paid leave would rarely continue beyond the first few months after the birth. More complex models would need to be estimated to take account of this. Despite this, these relationships between leave use and timing of return to work are useful to observe, especially as they relate to the proportion having returned to work after 12 or 18 months, when the majority of women's formal leave from work will have ended. In particular, comparisons can be made between those who used unpaid leave only, and those who used some or only paid leave, because it is expected that those women who used unpaid leave only did so because they were unable to access paid leave.

The other factors used in the multivariate analyses are described later, following a description of the techniques used.

Multivariate analyses

Two sets of logistic models were estimated to explore maternal return-to-work timing. The first set of regressions provided separate estimates for whether a mother had returned to work by 3, 6, 9, 12 or 18 months (inclusive) following the birth of a child. These estimates generated the cumulative probabilities of having returned to employment, moving further away in time from the birth. A second set of regressions provided estimates of the likelihood of returning to employment during specific periods following birth (when the child was aged 0-3 months, 4-6 months, 7-9 months, 10-12 months and 13-18 months), conditional on not having returned to work in an earlier period. These estimates inform us as to the hazard of returning during any of these periods.

This method was chosen over a more standard hazard model because initial exploration of these data revealed that the effect of pre-birth employment status and leave use on the likelihood of having returned to work or the hazard of return to work varied across the age of the child. Grouping the data and estimating separate models enabled these effects to be explored without the complication of numerous interaction terms.

For each dependent variable, two models were estimated. The first specification applied to all women, including pre-birth employment status in addition to the range of demographic variables. The second specification applied to mothers who were employees prior to the birth, including leave use, pre-birth employment status and other pre-birth job characteristics in addition to the demographic variables. Mothers who were self-employed prior to the birth were excluded, as some of the job characteristic details were not collected for these women.

Logistic models were appropriate because the variables of interest were binary, with zero indicating not returning (or not returned) and one indicating a return occurred (or had occurred). The analyses were adjusted for initial selection probabilities into the LSAC sample and the clustered nature of the data,⁴ using the "survey" commands in the Stata data analysis software.

To demonstrate the multivariate results, predicted values were calculated showing the predicted percentage returned at child ages of 3, 6, 9, 12 and 18 months, and the predicted hazard rate within the intervals. These predicted hazard rates were converted to predicted monthly hazard rates, so they were not affected by the number of months within each of these intervals. In each case, the variables other than the one under consideration were held at the sample mean (the mean of previously employed women for the latter model).

Other family characteristics and pre-birth employment variables

Control variables used in the analyses were family size, mother's education level, region and single-parent status. For partnered mothers, the partners' employment status and earnings (classified as < $500, $500-999, ≥ $1000 per week, gross) before the birth of the child was also included. Family size was operationalised, firstly by identifying the number of older siblings (grouped into none, one, two or more), and secondly, by using a yes/no variable to indicate if the mother was pregnant or had other children since the birth of the study child. The latter included mothers who had been pregnant in the 12 months prior to the survey (including at the time of the survey), even if the child had not yet been born. The regional variable classified respondents according to whether they lived in a highly accessible, moderately accessible or remote/very remote region (LSAC Project Operations Team, 2006).

Other pre-birth job characteristics were: size of business (1-19, 20-99, 100-499, 500 or more employees), own pre-birth earnings (< $500, $500-999, ≥ $1000 per week, gross), sector (public or private), usual hours worked (< 20, 20-34, ≥ 35 hours per week) and broad occupation group (manager/professional or other). A more detailed occupation grouping was explored but did not improve the explanatory power of the models.

The distribution of these variables is given in Appendix Table A1.

Results

Mothers' pre-birth employment and leave use

This section provides an overview of mothers' pre-birth employment status and use of leave as a background to the analyses of maternal return-to-work patterns.

Two-thirds of the women were employed in the 12 months prior to the birth of the study child and were most commonly working as permanent employees (43% of all women). Casual employees, contract employees and self-employed women made up 14%, 3% and 7% of the women respectively.

Table 1 shows that around three-quarters of women employed during pregnancy used some leave after the birth of the child, with 27% using no leave. Just 10% used paid leave only, 37% used both paid and unpaid leave, and another 25% used unpaid leave only. The majority of permanent employees used some leave after the birth, with just 12% using no leave. Most used paid leave (69%), including 56% who combined this with unpaid leave, and another 19% used unpaid leave only. Casual, contract and self-employed workers were much more likely to have used no leave (about 56%). When these previously employed mothers did take leave, they were more likely to take unpaid leave, although contract workers were more likely than casual or self-employed to have used paid leave.

Note: Percentages do not always add up to 100% due to rounding. 
Source: Parental Leave in Australia Survey, 2005

Paid leave was largely paid maternity leave, although some use was made of other types of leave, including holiday and long service leave (Table 2). Those who took paid leave only used, on average, a total of 20 weeks leave, including 12 weeks of paid maternity leave, 3 weeks of paid holiday leave, 3 weeks of paid long service leave and 1 week of other paid leave. Those who combined paid and unpaid leave took significantly more leave in total (45 weeks), although they used less paid leave than those using only paid leave. The average duration of unpaid maternity leave for these women was slightly higher (31 weeks) than for those women who used only unpaid maternity leave (27 weeks).

Source: Parental Leave in Australia Survey, 2005

These data provide some indication of the relationship between leave type and returning to work, with those using only paid leave likely to return to work considerably sooner than those who used unpaid leave. Based on these data, of those who were employed during pregnancy and took some leave, those who used both paid and unpaid leave were expected to have the slowest return to work.

Overall maternal return-to-work patterns

This section explores the timing of return to work for all mothers. Return-to-work timing is most simply viewed as the proportion of mothers who are in paid employment, according to the age of the child, measured in months (Figure 1). Very few mothers were in paid employment with a child aged less than one month (2% had returned at this time). At three months old, 11% of mothers had returned to work and at six months old, 22% had returned. By the time the child was 12 months old, 44% had returned and at 18 months, 54% had returned.

Figure 1: Percentage returned and hazard of return, all women

Source: Parental Leave in Australia Survey, 2005

For child ages of 2 to 8 months, of those mothers who had not yet returned to work, around 4% returned each month (the hazard of return to work). There was a slight increase in the hazard of returning at 9 to 10 months, and then a peak of around 7% when the child was aged 11 months. The hazard remained high for children aged 12 months (6%), but then fell in subsequent months. Returning to work when the child was aged 11 to 12 months is likely to be associated with the availability of one year's unpaid maternity leave for women who were in permanent employment before the birth.

Maternal return to work by pre-birth employment type

This section explores the relationship between pre-birth employment type and mothers' return-to-work timing. This is first done by examining the relationships between pre-birth employment type and the proportions returned to work, and the hazard of return to work according to child age, but without controlling for other characteristics. The results of the multivariate analyses are then presented, in which these same relationships are explored, after controlling for demographic variables.

Pre-birth employment type had a strong relationship with mothers' return-to-work patterns (Figure 2). At 18 months, the self-employed were the most likely to have returned (84%), followed by permanent (76%), contract (64%), casual (58%) and then not employed (13%). At 12 months, while fewer had returned, the differences by employment type were evident. At 9 months and earlier, however, there were far smaller differences between those who were previously permanent, casual or contract.

Figure 2: Percentage returned and hazard of return, by pre-birth employment type

Note: To simplify the hazard rate graph, contract workers were excluded. Their hazard rates were similar to those of casual workers, except in the first three months when contract workers had hazard rates similar to permanent workers. 
Source: Parental Leave in Australia Survey, 2005

The self-employed, albeit a small percentage (7% of the women), were more likely to return to work than others, especially while the infant was very young. They had a very high hazard of returning to work when the child was aged less than 3 months (53% had returned by 3 months).

In the first 3 months after the birth, those who were previously casual were more likely than the previously permanent (or contract) employees to have returned to work. The difference was, however, not large and, except for this effect, the return-to-work hazard was fairly similar for permanent, casual and contract employees while the child was less than 9 months old. After this time, the hazard of return to work increased for those who were previously permanent employees, with a peak at 11 to 12 months, and the return-to-work patterns of previously permanent and other employees diverged due to a much faster return to work among previously permanent employees. The more constant hazard rates of previously contract or casual workers meant that their return to work continued to increase more gradually as the child got older.

The cumulative proportion returned and the hazard rates were analysed using multivariate techniques to explore the effects of employment type, after controlling for differences in demographic variables (estimation results are given in Appendix Tables A2 and A3). Predicted values by pre-birth employment type were then calculated, holding other variables constant at the survey mean. These values are shown in Figure 3. Clearly, this process smooths out some of the month-to-month variation, but many of the relationships are still evident. These results (predicted values and the odds ratios) confirm the following results:

• Mothers who were not employed during pregnancy had the lowest hazards of return at all ages, and were therefore the least likely to have returned at all child ages.
• Mothers who were self-employed during pregnancy were much more likely than others to return to work in the first 3 months. Self-employed mothers' hazard of return to work fell in subsequent months, such that at 7-9 and 10-12 months it was lower than women who had been employed in a permanent job. However, because of the high proportion returning initially, the cumulative percentage returned was significantly higher for these women during the infants' first year.
• Casual workers also had a somewhat higher hazard of return in the first 3 months than permanent workers. Their hazard of return declined after this and was lower than that of permanent workers at 7-9 months and 10-12 months. As a consequence, the cumulative proportion returned increased more sharply for permanent workers than for casual workers. Relative to permanent workers, contract workers also had a slightly lower hazard of return to work at 10-12 months and 13-18 months.
• At 10-12 months, mothers who had been in a permanent job before the birth had a much higher hazard of return to work than others, resulting in a large increase in the proportion returned during this time. As previously noted, this is likely to be related to the availability of one year's unpaid maternity leave for these women.

Figure 3: Predicted percentage returned and hazard of return, by pre-birth employment type

Note: Predicted values were calculated after adjusting for differences in demographic and family factors, and other job characteristics. These variables were set at the survey mean.

Source: Parental Leave in Australia Survey, 2005

Maternal return to work by use of leave

This section considers the relationships between use of leave and maternal return to work. Bivariate relationships are examined first, then the multivariate analyses. These analyses were limited to mothers who were employees, that is, employed but not self-employed prior to the birth. About one-quarter of these women used no leave, another quarter used unpaid leave only, while 41% used both paid and unpaid leave and just over 10% used paid leave only.

Figure 4 shows maternal return to work by leave use. For those who took no leave, the highest hazard of return to work was in the first two months, as some of these women returned to work immediately or reported that they did not stop working. However, following this, the hazard of return was very low, resulting in a lower cumulative proportion returned, with 52% returned by 18 months. This was still considerably higher than the percentage returned among those that were not employed prior to the birth (13%).

Figure 4: Percentage returned and hazard of return, by leave use

Source: Parental Leave in Australia Survey, 2005

Women who used both paid and unpaid leave had a very low initial hazard of return to work. They were considerably less likely than women who used only paid leave or only unpaid leave to return to work when the child was aged less than 10 months old, as seen in the hazard of return and the cumulative percentage returned. However, at 10-12 months old, the hazard of return to work was high for these women and it was at this point that these women caught up in their return to work, such that the cumulative percentage returned was even higher than those who took only unpaid leave. By 18 months, 80% had returned to work.

Those who took only paid leave or only unpaid leave were fairly similar, although those using only paid leave were, across the 18 months, more likely to have returned to work, and to have returned sooner. These women had the highest proportion returned (83% by 18 months), to around the same level as the self-employed women. Nevertheless, those who took only unpaid leave also had a high proportion returned (77% by 18 months).

The hazard rate of return was highest for those who took leave, at around 11 to 12 months, coinciding with the completion of one year's maternity leave. However, those who took only paid leave also showed a peak at around 3 months, which probably coincided with an entitlement to 12 weeks paid maternity leave.

Multivariate analyses were used to determine whether these associations remained after allowing for other differences in characteristics (Appendix Tables A4 and A5). In addition to the demographic variables, these analyses took into account job characteristics prior to the birth. The predicted percentage returned and the predicted monthly hazard rates by leave use, calculated by holding the demographics and job characteristics at the survey mean, are shown in Figure 5.

Figure 5: Predicted percentage returned and hazard of return, by leave use

Note: Predicted values were calculated after adjusting for differences in demographic and family factors, and other job characteristics. These variables were set at the survey mean.

Source: Parental Leave in Australia Survey, 2005

These results show the following:

• Compared to those who took leave, those who took no leave had a lower hazard of return to work and a significantly lower cumulative proportion returned to
• work at most child ages. The exceptions were in the first three months, when those who used paid and unpaid leave had rates of return to work as low as those who used no leave. Also, at 12 to 18 months, the hazard rates of return did not vary according to whether the mothers took leave or the type of leave used, although there were differences in the cumulative proportion returned by this time.
• In the first 3 months after the birth, among those who took leave, those with the highest hazard of return to work were those who used only paid leave. Those who took only unpaid leave had a lower hazard of return to work, but it was higher than those who took a combination of paid and unpaid leave. Between 3 and 6 months, then between 6 and 9 months, the differences in the hazard of return to work were considerably smaller by type of leave, although in the younger age bracket, women who took only paid leave had a somewhat higher hazard of return to work than those who combined paid and unpaid leave. By 9 months, then, more of those mothers who had used only paid leave had returned than mothers who had used some unpaid leave.
• Between 10 and 12 months, the hazard rate of return was highest for those using paid and unpaid leave. As shown in Figure 5, this resulted in an increase in the cumulative proportion returned, and by 12 months, the differences by leave type were smaller. Still, those who took only unpaid leave were somewhat less likely to have returned than those who took only paid leave.
• As mentioned above, between 12 and 18 months, the hazard of return to work did not vary significantly with type of leave used, so the relative rates of return remained much as they were at 12 months. That is, women who had used only paid leave remained the most likely to have returned to work. These women were significantly more likely to have returned to work than those who had used only unpaid leave, although the difference between these groups was not large. In between these groups, were those who used a combination of paid and unpaid leave. These women did not differ in their likelihood of having returned to work from either group. That is, among those who used paid leave, no significant differences were apparent according to whether this leave had been used in conjunction with unpaid leave, and among those who used unpaid leave, no significant differences were apparent according to whether this leave had been used in conjunction with paid leave.

With the inclusion of leave use in the models, the coefficients for casual or contract (versus permanent) employment reduced in size and significance, indicating that these associations, discussed in the previous section, were in part related to the different patterns of leave use across these employment types, as was shown in Table 1.

Maternal return to work by demographic and other job characteristics

This section considers the relationships between maternal return to work and demographic variables and other job characteristics that were included in these analyses as control variables. While not the main focus of this paper, it is worth pointing out some of the findings, as they also contribute information about maternal return to work. This section goes on to present some additional analyses that were undertaken in order to explore why the demographic variables often had quite weak relationships with maternal return to work.

Level of education had a strong association with maternal return to work (Appendix Tables A2 to A5). Generally, higher education was associated with a greater hazard of return. The exception was for while the child was aged up to 3 months, when those with a certificate or diploma, not those with a bachelor degree or higher, had the greatest likelihood of return. After controlling for job characteristics, the education effects became slightly weaker, but were still evident. Other differences according to family type (couple versus single), partner's work/income status, child's birth order, presence of younger sibling (or mother pregnant) and region, were small and often not significant, after other variables were held constant.

Looking at pre-birth job characteristics, the size of the business had a significant effect, in that women were more likely to return to work while the child was aged up to 3 months if they worked in a business with fewer than 20 employees. Employees from larger businesses (500 or more employees) were the most likely to return between 10 and 12 months. Other pre-birth job characteristics had very slight effects.

A possible reason for the lack of effect of many demographic variables and job characteristics is the strong association between these variables and both leave use and employment type (Appendix Tables A6 to A8). For example, mothers were more likely to be employed during pregnancy and more likely to be employed in a permanent job if they had a higher level of education, if they were pregnant with their first child and if they had a partner with medium to high earnings (Appendix Table A6).

To investigate whether the inclusion of employment and leave use data was masking the associations with personal or family characteristics, additional models were tested in which all pre-birth employment information and leave use data were excluded. Appendix Table A9 compares coefficients from Table A2 and these new models, for the proportions returned by 12 and 18 months. Birth order had different effects depending on whether pre-birth employment was included. When excluded, those with more children were less likely to return to work. However, controlling for pre-birth employment removed this association, since the likelihood of being employed during pregnancy was very strongly associated with birth order (Appendix Table A6). The effects of having a not-employed partner or being a single parent were stronger in the model that did not control for pre-birth employment (these mothers being less likely to have returned at 12 or 18 months, compared to other mothers). Education also had a stronger effect. Again, these variables were associated with the likelihood of being employed during pregnancy. Comparisons of the amount of variation explained by these models also showed that pre-birth employment was an important predictor of return to work. So, while the return-to-work models showed that pre-birth employment and leave use were key variables in explaining mothers' return to work, other family characteristics were important through their association with these variables.

Discussion

According to the PLAS, 44% of mothers had returned to work within 12 months after the birth of a child. In the month the birth occurred, very few returned to work, and in the following months there was a gradual return to work, up to the age of 11 to 12 months, which were the peak ages of return to work. Consistent with previous research, there was variation in the timing of return to work according to a range of maternal and family characteristics.

An unsurprising finding was that being employed prior to the birth was a very strong predictor of post-birth employment (Baxter, 2005; Baxter et al., 2007; Glezer, 1988; Hofferth, 1996; Joesch, 1994). A new finding was that being self-employed prior to the birth was related to a faster return to work. The early return to work of self-employed mothers may reflect their lack of access to any paid leave, and their desire or need to keep their business going. Another factor, likely to be related to mothers' ability to manage an early return to work, is that these self-employed mothers often work very short hours, and the hours are often flexible (Baxter & Gray, 2006; Baxter et al., 2007).

Over the longer term, fewer casual (and contract) workers had returned to work relative to permanent employees, since they did not experience the peak in return to work at 10 to 12 months that permanent workers experienced. This may have been because these employees did not have access to the 12 months unpaid leave that many permanent employees were probably returning from at this time. It is possible - although not able to be tested with these data - that extending 12 months unpaid leave to more casual workers might increase the proportion returning to work when children are around one year old.

The majority of women were permanent employees prior to the birth, and apart from the self-employed women, these women were the most likely to return after a year or so. Most of these women would have had access to one year's unpaid maternity leave (Whitehouse et al., 2007). Even so, a considerable proportion of these women did not use a full one year of unpaid leave, with 40% of women previously in permanent jobs returning by 9 months, for example. An analysis of reasons for return, and other factors, would be needed to understand the reasons for this, but it is likely that financial reasons contribute, especially among mothers who did not have much paid leave.

The higher rate of return to employment of permanently employed women appears to have been related to their use of leave, as these women were much more likely than others to have used any leave, but in particular, to have used paid leave or a combination of paid and unpaid leave.

Using leave was related to higher rates of return to employment, although there were different effects over the first 12 months. In particular, in the first 3 months, those who used a combination of paid and unpaid leave were less likely to return to work than other leave-users, no doubt because this combination of leave enabled them to take a longer absence. These women had the highest rate of return to work between months 10 and 12. From this time, mothers who used a combination of paid and unpaid leave did not have significantly different rates of return to work to those who used only paid leave. Across all months, those who used only paid leave tended to have a higher rate of return to work than those who used only unpaid leave, although by 18 months, the differences were quite small. That is, taking only paid leave rather than only unpaid leave was associated with a slightly higher rate of return to employment following having a birth. However, these differences were far smaller than those between women who did and did not take leave, and between women who were and were not employed during pregnancy.

Women who took no leave were quite different to other previously employed women. Some of these women went back to work straight after the birth, but of those who did not, in subsequent months they were less likely to return than other previously employed mothers. By 9 months, a smaller proportion had returned to work compared to those who took some leave. Overall, then, use of leave was associated with a higher proportion returning and a faster return to work, consistent with the findings of Glezer (1988). This is not really surprising, though, since the "no leave" category included those who left their job knowing that they wanted to take longer than one year away from work.

It is problematic to draw conclusions about how improved access to maternity leave might result in changes to the timing of return to work, as these data have been limited by the availability of data on use of leave, rather than entitlement to leave. It is not clear what proportion of those who used no leave had an entitlement to leave but elected not to take it. However, it seems plausible to assume that women who used only unpaid leave did so because they had no paid leave entitlement. If these women had also had paid leave, it is possible that their rate of return to work would have increased a little, to be comparable with those who used a combination of paid and unpaid leave, or who used paid leave only. It is unlikely that a dramatic increase in maternal employment following the birth would result from such a change. It is not possible to determine, however, whether some of those women who used no leave would take leave and return to work sooner, if availability of parental leave was different.

Most other job characteristics had a small effect on women's return to work. One relationship that was clear was women who worked for small businesses had a faster return to work, even after excluding the self-employed. This is consistent with the ABS findings that women working in small businesses are the least likely to use maternity leave (ABS, 2007). Those who had been employed in larger organisations were the least likely to return in the first 6 months, but by 12 months were just as likely to have returned as other employees, given the higher probability of returning between 10 and 12 months.

A number of variables controlled in this analysis, including family type and partner's income, did not have a strong relationship with return-to-work timing when pre-birth employment was included in the analyses. However, the relationships between these variables and pre-birth employment (and leave use) were strong, and therefore had an association with return to work in an indirect way.

At this point, it is worth reiterating one limitation of this analysis - that this sample was biased towards those more connected to the labour market. The results, therefore, cannot be generalised to the entire female population. This is more of problem in analysing the return-to-work behaviour of those who were not employed during pregnancy, or those with lower levels of education, given the lower response rate for these women.

These analyses could only consider a limited set of explanatory variables and maternal return to work. The extent to which the family relied on government support as a main or important source of income was one factor that was not included. As in any country in which government support is means-tested, the withdrawal of support as other income increases can lead to high effective marginal tax rates for some families. This can mean that the financial incentive for mothers to work can be limited, particularly in low-income families (Beer, 2003; Buddelmeyer, Dawkins, Freebairn, & Kalb, 2004; Buddelmeyer, Freebairn, & Kalb, 2006; Toohey & Beer, 2004). Such effects are likely to be taken into account by mothers when considering their return-to-work options. Similarly, another factor that is likely to be important to parents that was not taken into account in this analysis, is the availability and cost of non-parental child care.

This analysis has only focused on whether or not a return to work occurred after the birth of a child. A more detailed examination of factors associated with a mother's return to work might look more closely at the nature of the return to work. This would include the hours worked on return, whether the return was to the same job or occupation as before the birth, and whether a change was made between job types - for example changing to self-employment. While beyond the scope of this paper, these details would further our understanding of how childbearing interacts with women's involvement in the labour market.

Data from the Parental Leave in Australia Survey could be used to examine further the reasons for women's return to work, and it is expected that this will be the subject of future analyses.

Conclusion

This paper uses recent Australian data to provide analyses of maternal return to work in the 18 months after childbirth. In addition to providing an overview of these data, this paper contributes to our understanding of how different features of the labour market and of maternity leave are associated with patterns of maternal employment in these months when children are young.

A clear finding was that women who were employed in the year before the birth had a higher rate of return to work than those who were not employed. Further, pre-birth employment status had a strong association with return-to-work timing, with women who were self-employed or in permanent jobs before the birth having the highest rate of return to work by 18 months after the birth. Self-employed women in particular had a very high rate of return straight after the birth.

Some of the patterns of return to work were related to differences in the use of paid and unpaid leave, but much of this was due to the lower rate of return to work of those who used no leave. By 18 months, those who used some leave had higher rates of return to work than other women. Also, women who used only paid leave had slightly higher rates of return to work than those who used only unpaid leave. Among those who used some paid leave, there was no significant difference in the likelihood of having returned according to whether or not this leave was combined with some unpaid leave.

This analysis provides more detailed insights to maternal employment following childbirth than is possible from conventional analyses of cross-sectional data. By identifying those women who have a slower return to work - for example, those who were not employed during pregnancy or who were employed but took no leave - we can identify those women who might face longer-term difficulties in the labour market. Also, by identifying those who return to work early - for example, those who return to self-employment - we can consider whether these women might face particular maternal or child health issues because they return to work soon after the birth. The comprehensive information on child and maternal characteristics, along with the longitudinal nature of the LSAC data will allow follow-up of some of these issues.

References

• Alexander, M., & Baxter, J. (2005). Impacts of work on family life among partnered parents of young children. Family Matters, 72, 18-25.
• Arun, S. V., Arun, T. G., & Borooah, V. K. (2004). The effect of career breaks on the working lives of women. Feminist Economics, 10(1), 65-84.
• Australian Bureau of Statistics. (2001). Census of Population and Housing: Household sample file (basic data). Confidentialised unit record file (CURF), 2001 (Cat. No. 2037.0.30.001). Canberra: Author.
• Australian Bureau of Statistics. (2006). Pregnancy and employment transitions, Australia, 2005 (Cat. No. 4913.0). Canberra: Author.
• Australian Bureau of Statistics. (2007). Australian social trends 2007 (Cat. No. 4102.0). Canberra: Author.
• Australian Institute of Family Studies. (2005). Growing Up in Australia: The Longitudinal Study of Australian Children annual report. Melbourne: Author.
• Baxter, J. (2005). Mothers' employment transitions following childbirth. Family Matters, 71, 11-17.
• Baxter, J., & Gray, M. (2006). Paid work characteristics of mothers with infants. Family Matters, 77, 34-41.
• Baxter, J., Gray, M., Alexander, M., Strazdins, L., & Bittman, M. (2007). Mothers and fathers with young children: Paid employment, caring and wellbeing (Social Policy Research Paper No. 30). Canberra: Department of Families, Community Services and Indigenous Affairs.
• Beer, G. (2003). Work incentives under A New Tax System: The distribution of effective marginal tax rates in 2002. Economic Record, 79(Special), S14-S25.
• Beggs, J. J., & Chapman, B. J. (1988). The foregone earnings from child-rearing in Australia (Discussion Paper No. 190). Canberra: Centre for Economic Policy Research, ANU.
• Berger, L. M., Hill, J., & Waldfogel, J. (2005). Maternity leave, early maternal employment and child health and development in the US. Economic Journal, 115(501), F29-F47.
• Breusch, T., & Gray, E. (2004). New estimates of mothers' forgone earnings using HILDA data. Australian Journal of Labour Economics, 7(2), 125-150.
• Brooks-Gunn, J., Han, W.-J., & Waldfogel, J. (2002). Maternal employment and child cognitive outcomes in the first three years of life: The NICHD study of early child care. Child Development, 73(4), 1052-1072.
• Buddelmeyer, H., Dawkins, P., Freebairn, J., & Kalb, G. (2004). Bracket creep, effective marginal tax rates and alternative tax packages. The Mercer: Melbourne Institute Quarterly Bulletin of Economic Trends, 1.04, 17-28.
• Buddelmeyer, H., Freebairn, J., & Kalb, G. (2006). Evaluation of policy options to encourage welfare to work. Australian Economic Review, 39(3), 273-292.
• Dex, S., Joshi, H., Macran, S., & McCulloch, A. (1998). Women's employment transitions around child bearing. Oxford Bulletin of Economics and Statistics, 60(1), 79-98.
• Edwards, R. (2006). Maternity leave and the evidence for compensating wage differentials in Australia. Economic Record, 82(258), 281-297.
• Glezer, H. (1988). Maternity leave in Australia: Employee and employer experiences. Report of a survey. Melbourne: Australian Institute of Family Studies.
• Goward, P. (2002). A time to value: Proposal for a national paid maternity leave scheme. Sydney: Human Rights and Equal Opportunity Commission.
• Gray, M., Qu, L., de Vaus, D., & Millward, C. (2002). Determinants of Australian mothers' employment: An analysis of lone and couple mothers (Research Paper No. 26). Melbourne: Australian Institute of Family Studies.
• Hofferth, S. L. (1996). Effects of public and private policies on working after childbirth. Work and Occupations, 23(4), 378-404.
• Hyde, J. S., Klein, M. H., Essex, M. J., & Clark, R. (1995). Maternity leave and women's mental health. Psychology of Women Quarterly, 19, 257-286.
• Joesch, J. M. (1994). Children and the timing of women's paid work after childbirth: A further specification of the relationship. Journal of Marriage and the Family, 56(2), 429-440.
• Joesch, J. M. (1997). Paid leave and the timing of women's employment before and after birth. Journal of Marriage and the Family, 59(4), 1008-1021.
• LSAC Project Operations Team. (2006). The Longitudinal Study of Australian Children data users guide, version 1.3. Melbourne: Australian Institute of Family Studies.
• Lyness, K. S., Thompson, C. A., Francesco, A. M., & Judiesch, M. K. (1999). Work and pregnancy: Individual and organizational factors influencing organizational commitment, timing of maternity leave, and return to work. Sex Roles, 41(7-8), 485-508.
• Macran, S., Joshi, H., & Dex, S. (1996). Employment after childbearing: A survival analysis. Work, Employment and Society, 10(2), 273-296.
• McGovern, P., Dowd, B., Gjerdingen, D., Moscovice, I., Kochevar, L., & Murphy, S. (2000). The determinants of time off work after childbirth. Journal of Health Politics Policy and Law, 25(3), 527-564.
• Miller, C. F. (1993). Part-time participation over the life-cycle among married-women who work in the market. Applied Economics, 25(1), 91-99.
• Organisation for Economic Co-operation and Development. (2007). Babies and bosses: Reconciling work and family life. A synthesis of findings for OECD countries. Paris: Author.
• Ondrich, J., Spiess, C. K., Yang, Q., & Wagner, G. G. (1999). Full time or part time? German parental leave policy and the return to work after childbirth in Germany. Research in Labor Economics, 18, 41-74.
• Polachek, S. W., & Sieber, W. S. (1993). The economics of earnings. Cambridge: Cambridge University Press.
• Productivity Commission. (2008). Inquiry into Paid Maternity, Paternity and Parental Leave (Productivity Commission Issues Paper). Canberra: Author.
• Pylkkänen, E., & Smith, N. (2003). Career interruptions due to parental leave: A comparative study of Denmark and Sweden (Social, Employment and Migration Working Papers No. 1). Paris: OECD.
• Rønsen, M., & Sundström, M. (2002). Family policy and after-birth employment among new mothers: A comparison of Finland, Norway and Sweden. European Journal of Population, 18(2), 121-152.
• Soloff, C., Lawrence, D., & Johnstone, R. (2005). Sample design (LSAC Technical Paper No. 1). Melbourne: Australian Institute of Family Studies.
• Swan, W., Gillard, J., & Macklin, J. (2008, February). Productivity Commission to investigate paid maternity leave (Media statement). Canberra: Australian Labor Party.
• Toohey, M., & Beer, G. (2004). Financial incentives to work for married mothers under a new tax system. Australian Journal of Labour Economics, 7(1), 53-69.
• Waldfogel, J., Higuchi, Y., & Abe, M. (1999). Family leave policies and women's retention after childbirth: Evidence from the United States, Britain, and Japan. Journal of Population Economics, 12(4), 523-545.
• Whitehouse, G., Baird, M., & Diamond, C. (2005). Parental Leave in Australia Survey: Funded through Australian Research Council Linkage Project LP0453613 and conducted as a nested study of Wave 1.5 of the Longitudinal Study of Australian Children.
• Whitehouse, G., Baird, M., & Hosking, A. (2007). Maternity leave in Australia: Patterns of use and return to work. In E. Rasmussen (Ed.), Diverging employment relations patterns in Australia and New Zealand: CD-ROM proceedings of the 21st Conference of the Association of Industrial Relations Academics of Australia and New Zealand (AIRAANZ). Refereed papers. Auckland: University of Auckland, AIRAANZ.

Appendix

Legend: # p < .1, * p < .05, ** p < .01, *** p < .001 
Note: Below each odds ratio, the 95% confidence interval around this estimate is shown in brackets. For the categorical explanatory variables, the omitted category is shown in brackets. It was not possible to derive a measure of pseudo-R-square with survey commands, so to estimate model fit, this value was derived by estimating the same model without the survey commands. 
Source: Parental Leave in Australia Survey, 2005

Legend: # p < .1, * p < .05, ** p < .01, *** p < .001 
Note: Below each odds ratio, the 95% confidence interval around this estimate is shown in brackets. For the categorical explanatory variables, the omitted category is shown in brackets. The "hazard of returning" is the probability of returning in this age group, given a return has not happened before this, so is only calculated for those who have not yet returned. It was not possible to derive a measure of pseudo-R-square with survey commands, so to estimate model fit, this value was derived by estimating the same model without the survey commands. 
Source: Parental Leave in Australia Survey, 2005

Legend: # p < .1, * p < .05, ** p < .01, *** p < .001 
Note: Below each odds ratio, the 95% confidence interval around this estimate is shown in brackets. For the categorical explanatory variables, the omitted category is shown in brackets. It was not possible to derive a measure of pseudo-R-square with survey commands, so to estimate model fit, this value was derived by estimating the same model without the survey commands. Excludes previously not employed and self-employed. 
Source: Parental Leave in Australia Survey, 2005

Legend: # p < .1, * p < .05, ** p < .01, *** p < .001 
Note: Below each odds ratio, the 95% confidence interval around this estimate is shown in brackets. For the categorical explanatory variables, the omitted category is shown in brackets. The "hazard of returning" is the probability of returning in this age group, given a return has not happened before this so is only calculated for those who have not yet returned. The pseudo-R-square could not be derived in the analysis with survey commands, so to estimate model fit this value was derived by estimating the same model without the survey commands. Excludes previously not employed and self-employed. 
Source: Parental Leave in Australia Survey, 2005

Source: Parental Leave in Australia Survey, 2005

Note: Excludes women not employed during pregnancy and self-employed during pregnancy. 
Source: Parental Leave in Australia Survey, 2005

Note: Excludes women who were not employed or self-employed prior to the birth. 
Source: Parental Leave in Australia Survey, 2005

Legend: # p < .1, * p < .05, ** p < .01, *** p < .001

Note: Below each odds ratio, the 95% confidence interval around this estimate is shown in brackets. For the categorical explanatory variables, the omitted category is shown in brackets. It was not possible to derive a measure of pseudo-R-square with survey commands, so to estimate model fit, this value was derived by estimating the same model without the survey commands.

Source: Parental Leave in Australia Survey, 2005 Source: Parental Leave in Australia Survey, 2005 Source: Parental Leave in Australia Survey, 2005 Source: Parental Leave in Australia Survey, 2005