3 Descriptive evidence on post-childbirth employment, job mobility and earnings of Canadian mothers

View the most recent version.

3.1 Post-childbirth employment

The birth of a child increases the benefits for the mother to stay at home, and if the mother stays home for an extensive period, her propensity to work in the future will likely be reduced, since a long career interruption can hinder her competence on the job or her chance to find a new job. It is thus desirable to investigate both the short- and the long-term employment effects of childbirth.¹²

We use the percentage of mothers who return to work in the first post-childbirth year to measure the short-term employment effect of childbirth. We refer to this percentage as the short-term post-childbirth employment rate. It simply answers the following question: of all mothers who give birth in year t, how many of them are employed in year t+1? Similarly, we use the employment rates in the other post-childbirth years to measure the longer-term effects of childbirth on the employment of the mothers. Figure 1 describes the employment rates for Canadian mothers in the first, second and the third post-childbirth years. Column 1 of Table 2 and Column 1 of Appendix Table A.2 contain the employment rates for the other post-childbirth years and for those women in the comparison group.¹³

Figure 1
Conditional employment rates for mothers and other women

Three observations can be made from Figure 1. The first observation is that both the long- and the short-term employment rates of Canadian mothers were consistently lower than those of their non- mother counterparts. For example, the short-term employment rate of the 1984 cohort of mothers was 84%, 13 percentage points below that of their non-mother counterparts; while the employment rate of the 2001 cohort of mothers in 2004 (the third post-childbirth year) was 87%, the corresponding employment rate of their comparison group was 91%. As mentioned before, the birth of a child increases the marginal costs and reduces the marginal benefits of working, and hence it is not surprising to observe that the post-childbirth employment rates of Canadian mothers were generally lower than their non-mother counterparts.

The second observation from Figure 1 is that the short-term post-childbirth employment rates of successive cohorts of Canadian mothers increased from the mid-1980s to the end of the 1990s and then started to decline in the early 2000s. For example, for the 1984 cohort of mothers, the employment rate of mothers in the first post-childbirth year was 84%. The employment rate reached an all-time high of 91% for the 1999 cohort of mothers, and then it dropped to the 87%-to-88% level for the early 2000s cohorts of mothers.

The above observation seems to suggest a non-linear relationship between the short-term post- childbirth employment rates and the generosity of the job-protected maternity system (see the Appendix, Evolution of the job-protected maternity leave system in Canada). When the protection is short, the employment rates of the new mothers in the first post-childbirth year are low. When the protection is extended moderately, the post-childbirth employment rates increase. But when the system is substantially extended to more than a year, the short-term post-childbirth employment rates decline.¹⁴

The non-linear relationship can be explained as follows. When the job-protected leave was relatively short, such as 17 weeks, some mothers were simply not able to physically recover from the ordeal of childbirth, others might have had difficulty in leaving their 17-week old children to somebody else's care. These mothers might have chosen to stay at home and give up their right to return to their pre-childbirth jobs. As a result, they might have had to look for new jobs when they were ready to work. But when the job-protected maternity leave becomes moderately long, as occurred during the 1990s in Canada, the above employment barriers would disappear or at least be subdued and, consequently, the post-childbirth employment rates would increase.

But when the system was extended to more than a year, it became feasible for some mothers to take the whole first post-childbirth year off and still retain their rights for the pre-childbirth jobs. For example, in 2000, about 29% of mothers were from Quebec, Manitoba and New Brunswick. Under provincial legislations, Quebec mothers were able to retain their pre-childbirth jobs for up to 70 weeks, while mothers from Manitoba and New Brunswick were able to return to their pre-childbirth jobs after 54 weeks. Thus, Quebec mothers who gave birth in the period from the second half of August to December, and mothers from Manitoba and New Brunswick who gave birth in the second half of December 2000, were all able to take the whole of 2001 off and return to their previous jobs by January 2002. Hence, the observed decline in the short-term employment rate for the 2000 cohort of Canadian mothers was very likely the result of some of these mothers taking more than a year off work. The same explanation applies to the declines of the short-term employment rates of the 2001-to-2003 cohorts of Canadian mothers.¹⁵

Table 2
Conditional employment and withdrawal rates of Canadian mothers

The third observation from Figure 1 is that the long-term post-childbirth employment rates of Canadian mothers increased from the mid-1980s and reached an all-time high in 1999, without any diversion. But the long-term employment rates of the early 2000s cohorts of mothers declined slightly. Since the employment rates of the corresponding comparison group also declined from the early 2000s, the levelling off in the long-term employment rates of the early 2000s cohorts of mothers seemed to be non-substantial.

To verify the above, we also use the long-term withdrawal rates by Canadian mothers to provide an alternative measure for the post-childbirth employment patterns for Canadian mothers. We have calculated the labour force withdrawal rates in the first three and five post-childbirth years.¹⁶ The results (Table 2, right panel) indicate that women who gave birth in the early 2000s were less likely to have withdrawn from the labour market during the post-childbirth years than

their counterparts in the mid-1980s had. For example, about 8% of mothers who gave birth in the mid- and late 1980s withdrew from the labour market in the first three post-childbirth years, but in the late 1990s and early 2000s, less than 6% of them withdrew from the labour market during the first three post-childbirth years.

3.2 Post-childbirth job mobility

The birth of a child brings out the work-family balance issues for the parents, particularly for the mothers. Although the tradition that female workers completely withdraw from the labour market upon giving birth has long gone, some mothers may still quit their jobs because of work-schedule inflexibility, commuting difficulties or the lack of childcare services. We have calculated the quit rate of new mothers in the first few post-childbirth years and the quit rates for the comparison group. The main results are plotted in Figure 2.

Figure 2 shows that the quit rates of the new mothers and the comparison group both fluctuated over time: mothers who had given birth during the downturn of the economy had lower quit rates than those mothers who gave birth during the booming years. But more importantly, when compared with non-mothers, the quit rates of mothers, in both the short- and the long-run, and in both the downturn and the upturn, were consistently lower, and the differences became more evident over time. For example, in the mid-1980s, the quit rates of new mothers in the first post-childbirth years were generally below those of the reference group by about 1 percentage point; by the early 2000s, the difference increased to more than 3 percentage points.

The observation that mothers had lower quit rates than other women can readily be expected, since the quit rates of mothers were measured over a group of women who returned to the labour market after having given birth. Those who had not yet returned to work were not part of the population from which the quit rate was calculated. Also, on average, it is not unreasonable to assume that mothers who returned to the labour market were those who had stronger labour market attachments, stronger career motivations, and/or more productive job-matches than those mothers who had not returned to work—some of whom may never have returned—particularly in the longer term. In other words, the quit rates for mothers are defined over a selected group of mothers who had a relatively strong labour market attachment, and hence, it is possible that their quit rates were below the average quit rates of the comparison group.

Figure 2
Post-childbirth quit rates

Standard human capital theory suggests that whether a mother returns to her pre-childbirth employer/job can make a significant difference in the labour outcomes, since returning to the same employer/job implies that a mother would lose little of her firm-specific human capital or job tenure, particularly if she returns to her pre-childbirth job relatively quickly. Hence, it is interesting to directly examine the proportion of new mothers who were working for their pre-childbirth employers during the post-childbirth years.

These results are plotted in Figure 3 for the first and the third post-childbirth years. Again, the likelihood to work for the same employer was affected by economic fluctuations for both mothers and other women, primarily because less outside opportunity was available in an economic downturn than in an upturn. However, the more revealing result is that, as indicated in Panel 1, before 1991, Canadian mothers were slightly less likely to work for the same employer one year later than other women; the difference was about 2 percentage points. But since 1991, the proportion of new mothers staying with the same employer rose to the same level as that of other women, and from 2001, new mothers have become somewhat more likely than other women to stay with the same employer.

In terms of the proportion of women staying with the same employer in the longer run, we have found that successive cohorts of new mothers were more likely to stay with the same employer than the reference group in the past 20 years, and the differences stayed relatively constant, as indicated by Panel 2 of Figure 3. One reason for this might be that mothers with young children were less mobile than women from the comparison group, perhaps the presence of young children has made job change undesirable; for example, some working mothers with young children need to make new childcare arrangements when changing employers/jobs, while for the comparison group—women without children—such a barrier does not exist.

Figure 3
Proportion of women working for the same employer

3.3 Earnings of Canadian mothers before and after childbirth

A simple measure of the effect of childbirth on the earnings of mothers is to compare the pre- and post-childbirth earnings of the mothers. This helps us to answer some interesting questions, such as to determine the immediate earnings drops that occurred to the mother when a child was born, how many years it took Canadian mothers to attain their pre-childbirth earnings, and how the earnings recovery process has evolved over the last 20 years.

The advantage of this approach is that we do not have to impose any strong sampling restrictions and hence we are able to work with a broad sample of Canadian mothers. The main disadvantage is that a simple comparison of the pre- and post-childbirth earnings does not reveal the true earnings effects of childbirth, since we do not know how the earnings of the mothers would have grown had they not have become mothers. Nevertheless, the comparison will provide us with some guidelines on the earnings effect of childbirth.

Table 3
Earnings changes (percent) relative to the pre-birth year (t-1)

Table 3 contains the changes in the average annual earnings for the 1984-to-2003 cohorts of Canadian mothers, where the percentages indicate how the earnings of the mothers changed in the year of childbirth and in the post-childbirth years, when compared with the earnings in the year before the child was born. The table suggests that the earnings drops for Canadian mothers were quite strong during the year of childbirth and in the first post-childbirth year, and that these drops tended to increase over time. During the 1980s, the birth of a child brought down the earnings of the mother by about 28% in the year of childbirth relative to her pre-childbirth earnings. This increased to the 30% level in the 1990s, and to about 33% after 2000. During the first post-childbirth years, the earnings drops relative to the pre-childbirth level were between 14% and 18% before 2001 and, since 2001, they have risen to the 37%-to-39% level.

The earnings drop in the year of childbirth and in the first year after childbirth were likely the results of the changing maternity leave and benefit systems. Prior to 1991, the maximum duration of the maternity leave benefits was 15 weeks. Therefore, those who gave birth within the first 37 weeks of a year would be able to collect the benefits during the year of childbirth. Assuming that the time of childbirth is uniformly distributed over the 52 weeks of a year, this group of mothers would then account for about 70% (37÷52=0.71) of mothers who gave birth within that year. If they all exhausted the maternity benefits (by collecting the Unemployment Insurance benefits for 15 weeks) and incurred no other earnings interruptions, they would lose about 30% of their potential earnings (15÷52=0.29). On the other hand, mothers who gave birth in the last 15 weeks of a year would incur less than 30% of an earnings drop during the year of childbirth, but they would incur some earnings drop by continuously taking their maternity leave and collecting the maternity benefits during the subsequent year (the first post-childbirth year).

Similar calculations can be made for other cohorts of mothers. In particular, the large earnings drops in the first post-childbirth year for the 2001-to-2003 cohorts of mothers can be easily understood because, from 2001, Canadian parents (mostly mothers) were able to receive Employment Insurance (EI) benefits for up to 50 weeks. Then, first, those who gave birth in the early part of a year were able to collect the benefits for almost a whole year, and therefore their earnings drops during the year of childbirth were now higher than those incurred by mothers who had given birth before 2001. Second, those who gave birth in the second part of the year could continue to collect the benefits for a significant part of the first post-childbirth year, and indeed for those who gave birth at the end of a year, they could rely on EI benefits for most part of the first post-childbirth year, and therefore, we observe higher earnings drops in the first post-childbirth year for the early 2000s cohorts of mothers.

Table 3 also suggests that the earnings of the 1990s and early 2000s cohorts of mothers recovered faster than those of the mid-1980s cohorts of Canadian mothers. For example, when we compare the earnings at t+5 (five years after childbirth) and t-1(one year before childbirth), it can be seen that for the mid-1980s cohorts of Canadian mothers, the t+5 earnings were only slightly above those of t-1. But for those mothers who gave birth after the mid-1990s, the t+5 earnings were generally higher than those in t-1 by 10% or more. The faster earnings recovery in the early 2000s seems to have been mainly due to the better labour market conditions observed since the late 1990s.

 

¹² A woman was defined as employed in a year if she had positive earnings from one or more paid jobs.

¹³ Notice that the horizontal axis of the figure represents different cohorts of mothers and non-mothers. It also measures time (year) implicitly. For example, in Panel 1, the 84% for the 1984 cohort of mothers indicates the employment rate of this cohort of mothers in 1985 (the first post-childbirth year).

¹⁴ This echoes the finding of Baker and Milligan (2005) that a moderate expansion of the job-protected maternity leave system does not, while a significant expansion of the system does, increase the time that new mothers spend at home.

¹⁵ We conducted probit analyses on both the short- and long-term employment of Canadian mothers by controlling age, cohort, employer size, previous earnings and province. None of these variables can explain the employment differences between mothers and other women.

¹⁶ We define a mother as a withdrawer in a three (five) post-childbirth year window if she did not receive any earnings in three (five) consecutive post-childbirth years. We do not define the short-term (such as one-year) withdrawal rates because a mother who did not work in the first post-childbirth year might have just been taking her job-protected maternity leave and hence was not withdrawing from the labour market.