1 Introduction

View the most recent version.

As a key family event, the birth of a child has important implications for the wellbeing of the mother, the child and the family. Economists are primarily interested in the effects of childbirth on the labour supply and the earnings of the mothers, and extensive studies have been conducted in these areas.¹ More recently, two strands of literature on childbirth and the labour-market outcomes of women draw particular interests: one focuses on the effect of maternity/parental-leave policies on the post-childbirth employment of the mothers; and the other concerns the 'family gap,' the pay or earnings difference between women who have children and childless women.

The two strands of literature are closely related. The birth of a child has an immediate negative effect on the employment of the mother because upon the birth of a child, the benefits of staying at home and the costs of working rise for the mother. Also, if she stays out of the labour market long enough, on account of, for example, extensive maternity/parental leaves, her human capital may depreciate or a productive job-worker match may disappear. As a result, her pay rate and earnings would be negatively affected, not only during the years immediately after the birth but also over a number of post-childbirth years.²

Available studies, however, tend to focus on the immediate or the short-term effects of childbirth, paying little attention to the long-term employment and earnings impacts of childbirth, often because of data limitations. For example, Baker and Milligan (2005) examine the post-childbirth employment effect of job-protected maternity/parental leave legislations in Canada up to four months using data from Statistics Canada's Labour Force Survey, while Marshall (1999) and Hanratty and Trzcinski (2005) examine the employment of mothers up to two years after childbirth. One recent study that pays attention to the long-term employment effect of childbirth in Canada is Skuterud (2008); but he focuses on married mothers who experienced one childbirth only. On the other hand, studies on the family gap typically estimate a single 'motherhood effect' on the wage or earnings of the mother and they rarely examine how the effect evolves along the post-childbirth years.

In contrast to previous work, we investigate the employment and job mobility patterns of Canadian mothers in both the short and the long run. We pay particular attention to the evolution of the short- and long-run post-childbirth employment over the past 20 years in Canada, during which the maternity related benefits and job protection legislations changed dramatically. Since these policy and legislations applied to employed women only, the underlying population of interest consists of women who had held a paid job for a certain time period before giving birth. Based on this population, we first establish a 'broad' sample of Canadian mothers who satisfied certain employment conditions that enabled them to enjoy the job-protected maternity/parental leaves and benefits.

Based on this broad sample, we found that (a) the short-term post-childbirth employment rates of Canadian mothers rose from the early 1980s to the year 2000, and have declined since then; (b) the long-term post-childbirth employment rates of Canadian women who had given birth in the early 2000s were much higher than that of their counterparts who had given birth in the mid-1980s; and, (c) Canadian mothers became less likely to quit or to change employer over time, when compared with their 'non-mother' counterparts.³

In the context of the family-gap literature, this study also examines the earnings trajectories of Canadian mothers. But, due to a number of data limitations, we focus on a 'narrow' sample of Canadian mothers—those who had a strong labour market attachment. Nevertheless, we are still able to make two contributions to the literature. On the one hand, several studies indicate that interruptions to one's career as a paid employee lead to earnings losses that can never be fully regained.⁴ Our result suggests that, at least with a group, and perhaps an increasingly important group, of mothers who have strong labour market attachment, the earnings losses associated with childbirth—a typical career interruption for women—decline over the post-childbirth years, and results from our most flexible model suggest that the impact disappears after seven years. On the other hand, it can be argued that the motherhood-earnings penalty is weak for mothers who have strong labour-force attachment.⁵ If this is the case, our result should help to gauge the lower limit of the motherhood-earnings penalty.

One feature of this study is that we borrow techniques from the program evaluation literature by establishing a comparison (or control) group of women for different cohorts of mothers (the treated group). This allows us to obtain the difference-in-differences estimates for the parameters of interest based on the counterfactual earnings for mothers—the earnings they would have received had they not have become mothers. With our data, we are able to estimate both the fixed-effects model— where individual specific intercept is allowed—and the fixed-trend model—where individual specific-earnings growth path is also allowed.

For the narrow sample of Canadian mothers, our results indicate that (a) the earnings drop for the mothers during the year of childbirth varied around 40%, and the maternity/parental leave benefits compensated for about half of this drop; (b) the negative effects of childbirth on the earnings of the mothers declined over the post-childbirth years and the result from the most flexible model (fixed trend) suggests that the effect disappeared after seven years; and (c) the result from the fixed-trend model also implies that the negative effects of childbirth for mothers who returned to their pre- childbirth employers were negligible beyond the second post-childbirth year.

The remainder of this paper is organized along the following lines. In Section 2, we discuss the dataset and establish the broad and narrow samples of mothers and their non-mother counterparts. In Section 3, we examine the post-childbirth employment of Canadian mothers, based on the broad sample. Section 4 introduces our main empirical framework, while Section 5 presents the estimates of the motherhood earnings penalty for the continuously employed mothers. In Section 6, we discuss the robustness of the results. A summary and the conclusions are contained in Section 7.

 

¹ Reviews of the early literature can be found in Killingsworth and Heckman (1986), Nakamura and Nakamura (1992) or Browning (1992).

² Browning (1992), for example, concludes that children aged up to 10 years old could be as costly as infants in terms of time demand to their parents.

³ The 'non-mother' women (the comparison or control group) were those who did not give birth during a specified period. They might have given birth before being observed or they may have given birth subsequently.

⁴ These include the work by Corcoran (1979), Jacobsen and Levin (1995), Mincer and Ofek (1982) and Stratton (1995). See also Phipps, Buirton and Lethbridge (2001).

⁵ For example, those with strong labour force attachment are often those who have strong human capital endorsement and tend to return to the labour market sooner rather than later after childbirth. As a result, they would experience less human capital losses associated with childbirth. See Shapiro and Mott (1994).