In 2015, the Department of the Navy tripled the amount of fully paid leave available to new mothers from 6 to 18 weeks. And though the Department of Defense forced the Department of the Navy to eventually scale back their leave to 12 weeks in order to create a consistent policy across the services, the large increase in maternity leave created a quasi-experiment on the impact of maternity leave on personnel. As part of a suite of studies on the effect of personnel policies on budgets and personnel inventories, sponsored by the Assistant Secretary of the Navy, we estimated the effect that the expanded maternity leave policy — combining the 18- and 12-week policies — had on the reenlistment of female sailors.

The results of this study indicate that longer paid maternity leave in the Navy increased the number of female sailors who reenlist, reducing what had been a small but persistent gender gap in reenlistment rates in a cost-effective manner. A concern surrounding the expansion of maternity leave was that it might result in fewer weeks worked, placing stress on Navy units. But in fact, we calculate that the extra weeks of work lost to new mothers were compensated by three times as many weeks of extra work from increased reenlistments.

Analyzing Navy maternity leave policy requires a different approach than that for civilian policy, but it comes with several advantages. Many studies of civilian maternity leave policies examine how likely women are to return to work after taking leave. Sailors, however, commit to fixed-length service obligations. So we instead look at how likely a sailor is to reenlist at the end of his or her initial service obligation. This also lets us look at whether additional maternity leave works as a reenlistment tool among women who do not yet have children but might be more likely to reenlist knowing that the longer maternity leave policy is available for future use. To our knowledge, no studies of civilian policy have addressed this topic. Finally, the new policy was quickly implemented, widely publicized and automatically made available to all sailors, making it an ideal experimental subject.

To estimate the policy’s effect, we used a difference-in-differences model. The model compares reenlistment rates of female sailors with those of male sailors without military spouses — who do not benefit, even indirectly, from the policy. The model also compares reenlistment rates before and after the policy went into effect. For about two years prior to the expansion of maternity leave, the reenlistment rates of male sailors were higher than those for female sailors. But after the new maternity leave policy was implemented, the gender gap in reenlistment disappeared. After controlling for sailors’ background characteristics and service histories, we find that the expansion of maternity leave increased the reenlistment rate of female sailors by 3.7 percentage points relative to that of male sailors.

Even female sailors without children seem to be encouraged to reenlist by better maternity leave. Their reenlistment rates increased by 3.6 percentage points, similar to those for female sailors overall. This result may suggest that sailors are thinking of future career and personal flexibility as much as benefits already received when making their reenlistment decision.

We can begin to quantify the value of those gains in dollars by comparing them to the Selective Reenlistment Bonuses (SRBs) the Navy pays to increase the reenlistment rate of sailors in particular skill specialties. For example, based on previous analyses we can estimate that to incentivize the same 3.7 percentage point increase in the reenlistment rate using bonuses, the Navy would have to offer Petty Officers Second Class (paygrade E-5) an extra $15,000 at reenlistment. Note, however, that SRBs affect male and female sailors roughly equally. Furthermore, SRBs are targeted to encourage reenlistment in occupations for which the Navy struggles to maintain sufficient manning and are not meant to address gender imbalances. Thus, SRBs are entirely appropriate and particularly efficient for addressing occupational shortages, but they are ill-suited for addressing a retention gender gap.

To return to the question of how weeks of work lost to longer maternity leave compare with weeks gained by higher reenlistment, we can compute back-of-the-envelope estimates. During fiscal year 2017, there were 1,236 births to four-year-obligation sailors. We estimate that compared with the old six-week maternity leave policy, these births generated approximately 8,500 additional weeks of leave under the new policy. During this period, 4,412 female sailors reached the end of an initial four-year obligation — the reenlistment decision point. If an additional 3.7 percent of these sailors reenlisted for an average duration of 3.5 years, this would translate to 164 sailors reenlisting for approximately 28,000 additional weeks. In short, by accepting a small additional loss of 8,500 weeks of work time by new mothers, the Navy could gain three times as many total work weeks from the sailors who might not have reenlisted without the increase in maternity leave. The difference is so large that even if these estimates were off dramatically and the reenlistment effect were up to three times smaller, there would still be a net gain in weeks worked. Furthermore, the additional reenlistments should increase sailors’ total sea duty; the length of maternity leave does not affect the amount of sea duty lost to pregnancy and childbirth.

While it remains to be seen whether the gains from this change in policy persist as it becomes the new status quo, it nevertheless provides proof of concept that non-monetary benefits may affect particular groups of sailors at least as effectively as cash bonuses. And while questions remain about maternity leave and workplace attachment in the civilian world, this study provides evidence that such policies may help employers retain their female employees prior to starting a family.


Daniel Leeds is a Research Analyst in CNA’s Resource and Force Readiness Division.