maternity and parental benefits in canada: are there...

Maternity and Parental Benefits in Canada: Are there Behavioural Implications?415

CANADIAN PUBLIC POLICY – ANALYSE DE POLITIQUES, VOL. XXVI , NO. 4 2000

Maternity and Parental Benefits inCanada: Are there BehaviouralImplications?SHELLEY A. PHIPPS

Department of EconomicsDalhousie UniversityHalifax, Nova Scotia

Cet article utilise des micro-données de l’Etude des Activités du Marché du Travail de 1988/89/90, pourexaminer certaines implications comportementales du régime canadien des allocations de maternité/parentales. Nous constatons d’abord que l’accessibilité des allocations n’influe pas de façon significativesur le comportement en ce qui concerne la fertilité et, deuxièmement, qu’il n’y a pas d’indication que lesfemmes adaptent leur comportement en matière d’emploi en vue de profiter de ces avantages. Nous exami-nons aussi qui peut bénéficier de ces allocations de maternité/parentales. Les mères adolescentes, les femmesn’ayant que peu d’instruction et celles qui connaissent des difficultés dans le marché du travail sont moinssusceptibles d’être éligibles. Etant donné qu’on ne constate pas de changement significatif des comporte-ments, il semblerait raisonnable de faciliter l’accès à ces avantages sociaux.

This paper uses micro data from the 1988/89/90 Labour Market Activities Survey to study some behaviouralimplications of the Canadian maternity/parental benefits system. We find, first, that fertility behaviour isnot significantly influenced by the availability of benefits, and, second, that there is no evidence that womenadjusted their labour-supply behaviour in order to gain access to benefits. We also examine who is potentiallyeligible for maternity/parental benefits. Teenaged new mothers, women with little education and thoseexperiencing difficulty in the labour market are less likely to be eligible. Given the evidence on lack ofsignificant behavioural response, it would thus seem reasonable to ease access to these benefits.

I n Canada, partial wage replacement is providedto new parents who take time away from a paid

job to care for their infants through the “maternityand parental benefits” component of the Employ-ment Insurance (previously UnemploymentInsurance) program. Thus, until 1997, applicants formaternity/parental benefits required 20 weeks ofpaid employment with either 15 hours per week orminimum weekly earnings. The switch from Unem-ployment Insurance (UI) to Employment Insurance(EI), means that applicants now require 700 hours

of paid employment — the equivalent of 20 weekswith 35 hours per week (with no minimum earningsrestriction).

Under either program, some new parents who arein the labour force will not satisfy the conditionsnecessary to receive benefits. In general, it is easierto receive regular UI/EI benefits than to receive ma-ternity benefits.1 Policy history clearly indicates thatthis reflects a concern about potential program abuseby new mothers (i.e., some women might enter the

416 Shelley A. Phipps


labour force prior to having a child simply in orderto gain access to benefits). When maternity benefitswere first added to the UI program in 1971, the 20-week entrance requirement was imposed as well asa “magic-10” rule requiring a maternity benefitsclaimant to prove that she had been in paid workten weeks prior to conception. The magic-10 rulewas abolished in 1984, though the extra work re-quirement (compared to most regular benefitsclaimants) has been maintained.2

But, should we be worried about such incentiveeffects when designing a maternity/parental benefitsprogram? The behavioural implications of theCanadian maternity/parental benefits program arenot well understood.3 We do not know, for exam-ple, whether fertility decisions are influenced by thisprogram,4 or whether women modify their laboursupply behaviour in order to establish eligibility. Ifbehavioural responses to the program are large, thenthe present program design makes sense. If, on theother hand, behavioural responses are minimal, thenit may be reasonable to change eligibility conditionsso that it is at least as easy to qualify for maternitybenefits as to qualify for UI/EI benefits (or perhapseven easier). It is important to remember that somepeople might genuinely be unable to obtain suffi-cient paid employment to ensure eligibility. This canbe particularly troublesome for Canadians in theirprime child-bearing years (i.e., aged less than 35years), who typically face higher unemploymentrates than older adults. Moreover, continued growthof non-standard employment (jobs with low weeklyhours and/or short durations) could mean that fewerand fewer new parents will be able to satisfy therequirements necessary to receive benefits.5

This paper focuses on two questions about po-tential behavioural responses to the Canadianmaternity/parental benefits system: Do some womenhave babies in order to take advantage of the pro-gram? and Do some women increase hours of paidlabour in order to gain access to benefits? Answersto these questions are important for sensible policydesign.

To address the first question, a probit model ofthe probability that a woman will have a baby isestimated in which the key explanatory variable is adummy variable which equals 1 if she was eligiblefor maternity benefits under UI, the program in placeat the time the data used for the analysis were col-lected. Results suggest that fertility is not sensitiveto the availability of maternity benefits. This doesnot seem unreasonable. It is perhaps more likely thata woman planning to have a baby might increaseher labour supply prior to the birth in order to enti-tle herself to benefits. To examine this issue, a probitmodel of the probability of increasing labour sup-ply enough (or more) than required to establisheligibility for maternity/parental benefits is esti-mated. Women who have just given birth or adopteda child6 are not significantly more likely to haveincreased their labour supply in the year precedingthe birth than other women, suggesting a lack oflabour supply response to the Canadian maternity/parental benefits system.

The final question addressed in the paper iswhether the switch from UI to EI is likely to haveaffected eligibility for maternity/parental benefits.The change from UI to EI means that every hour ofpaid work “counts” toward benefit eligibility. Thisshould improve access to benefits for some newparents with non-standard jobs, but the switch to anhours criterion means that workers with low weeklyhours must have more weeks of work in order to beeligible. For example, a woman working 20 weekswith 15 hours per week (300 hours) would have beenentitled to maternity benefits under the UI system;the same woman would require over twice as manyhours to qualify for maternity benefits under the EIsystem. Thus, the net effect of the program changefor access to maternity/parental benefits is not ob-vious, a priori.

This paper provides some initial estimates of theimpact. Given the lack of estimated behavioural re-sponse to maternity benefits under the UI systemreported in the first section of the paper, it is rea-sonable to assume that the change to EI will not have



significantly modified the labour market behaviourof new mothers. Hence, changes in eligibility asso-ciated with the program change are simplycalculated based on labour market history in the yearprior to the birth/adoption of the child.

Throughout, the paper uses data from the 1988/89/90 Labour Market Activities Survey (LMAS)panel. Although this data set is now ten years old,for the purposes of this study it offers two key ad-vantages: three years of data are available to enableus to study changes in labour market behaviour inthe year prior to having a baby; and the sample sizeis large enough to give us a reasonable sample ofnew mothers.7

Another caveat is that the paper focuses exclu-sively on women. Only women are eligible formaternity benefits, but both men and women areeligible for parental benefits (on the same terms).However, since only a small fraction of parentalbenefits claimants are men (e.g., never more than 6percent in 1999 — CANSIM, D739873), this paperfocuses on women.8

DATA

The 1988/89/90 Labour Market Activities Survey isa panel study of individual labour market experi-ence in Canada. Respondents were drawn from theLabour Force Survey sampling frame and were in-terviewed in three consecutive years (duringJanuary/February) about their previous year’s labourmarket experiences. In each year, respondents wereasked about the start and finish dates of each jobheld in the previous year (up to five jobs), the wageand non-wage characteristics of each job, the rea-sons for any interruption in employment, averagehours per day and days per week during each spellof employment. As well, basic demographic infor-mation was collected. (However, the LMAS containsextremely limited information about other familymembers.) Thus, this survey includes informationabout weeks of self-employment versus paid em-

ployment, hourly wage rates, and weekly hours —all of which are necessary to calculate potential eli-gibility for maternity/parental benefits. One majorlimitation of the LMAS from the perspective of thisresearch is that it is not possible to determine whichindividuals actually collect maternity benefits.

While it is not possible to identify all new birthswith certainty using the LMAS,9 a very good ap-proximation can be made by locating all familieswith an increase in the total number of childrenpresent from one year to the next and at least onechild in the 0-2 year age category; or an increase inthe number of children in the 0-2 year age categorywith no change in the number of children in the fam-ily (in case an older child leaves home). Animportant advantage of the data set is that we canobserve labour market behaviour for two years priorto the birth of the child, which then allows us toassess changes in behaviour immediately prior tothe birth which may be a result of the maternity/parental benefits program.

Table 1 presents basic sample characteristics. In1990, 1,080 women in the LMAS data set reportedthe birth/adoption of an infant (4 percent of allwomen aged 16 to 65). Thus, our weighted totalnumber of women estimated to have given birth/adopted in 1990 is 375,013 while CANSIM (label0144; matrix 0004) reports 405,486 births for thesame year, which does not seem unreasonable. (Ourestimates do not include multiple births.) The vastmajority of women who gave birth in 1990 were la-bour force participants in 1989 (884 women or 82percent).

A significant majority of women with infantswere in the 25 to 34 age range (61 percent). Womenaged 20 to 24 were the second most likely to have ababy in 1990 (27 percent). The largest number ofmothers with newborns have no other children (45percent of all new mothers; 52 percent of labourforce participants); the smallest number of mothershave 3 or more additional children (5 percent of allnew mothers; 2 percent of labour force participants).



TABLE 1Proportion of All Women with Babies in 1990 Eligible for Maternity/Parental Benefits – Weighted

All Women with All Women with a Baby in 1990a Baby in 1990 in Labour Force in 1989

All women 62.1 73.3Age

16-19 years 40.4 46.320-24 68.3 80.225-34 62.0 72.435-64 55.3 75.7

EducationLess than high school diploma 40.5 61.3High school diploma 65.5 76.9Some postsecondary 63.1 78.5Certificate/diploma/trade 69.7 75.0University degree 67.7 70.8

ProvinceAtlantic 61.2 71.9Quebec 61.5 81.1Ontario 64.5 68.7Manitoba 62.8 75.1Saskatchewan 64.0 73.5Alberta 61.3 74.4British Columbia 57.2 69.7

Underemployed in 1989 37.4 31.5Not Underemployed in 1989 64.6 78.2Occupation in 1989

Management/administrative – 77.0Professional – 81.0Clerical – 83.5Sales/services – 69.8Other – 62.6Farming/fishing – 36.0

Industry in 1989Primary – 45.3Manufacturing – 75.5Utilities – 94.0Retail/wholesale trade – 75.1Finance – 82.0Social services – 77.9Other services – 70.2Government – 87.1

No. of Children in 19890 77.6 79.91 58.4 68.72 42.9 66.13+ 10.6 30.7

Marital StatusMarried/living with partner 63.2 73.7Not married 54.2 70.5



A large majority of new mothers are married orliving with a partner (88 percent; versus 67 percentof all women). Patterns of occupation and industryresemble those for other women in the sample; newmothers are better-educated and live in higher-income households than women in general.

BASIC DESCRIPTIVE ANALYSIS

Since the economics literature on this subject is verysparse,10 this section of the paper provides a basicdescriptive analysis of the characteristics of thepopulation potentially eligible to receive maternity/parental benefits in Canada in 1990.11 Table 2 re-ports calculations of potential eligibility under theformer UI system for all new mothers12 as well asfor those who were in the labour force. Since weeksof self-employment do not count toward benefit eli-gibility, the nature of employment was checked foreach individual on a week-by-week basis. The esti-mates of el igibi l i ty for benefits emphasizedthroughout the paper are based on labour marketbehaviour in 1989.

In fact, eligibility for UI maternity/parental ben-efits should be based on employment history in the12 months preceding the claim. Thus, for example,if a woman began her maternity leave on 1 Decem-ber 1990, the eligibility period should be from 1December 1989 to 1 December 1990 (and not from1 January 1989 to 31 December 1989). To assesswhether the results reported in this paper are affectedby calculating eligibility based on 1989 work his-tory, we also determined the week in 1990 duringwhich a woman’s leave began, and then calculatedbenefit eligibility based on the 52 weeks precedingthe leave start-date. This was possible for 75 per-cent of the 884 women who had been in the labourforce in 1989 and had babies in 1990 because thesewomen actually reported time out of the labour forceat some point in 1990. Calculated eligibility statuswas the same using either method for all but 32women. Since qualitative results were not affected,we report only the results based on 1989 work his-

tory for everyone, noting any differences that seeminteresting.13

Table 1 indicates that 62.1 percent of all womenwho gave birth to (or adopted) a baby in 1990 couldpotentially have qualified for maternity/parentalbenefits based on 1989 employment history; 73.3percent of women who had a baby in 1990 and whowere in the labour force in 1989 could have quali-fied.14 While the LMAS does not report specificallyon receipt of maternity or parental benefits, we doknow whether the individual reported receipt of UI(which could include regular or sickness benefits aswell as maternity or parental benefits). Of thewomen we estimate to be potentially eligible forbenefits based on 1989 work history, 71 percent re-ported receipt of UI (a very crude estimate of benefittake-up).

Potential eligibility follows a definite age pattern.Presumably because many are still in school, teenagednew mothers were least likely to be eligible (only 40percent). New mothers aged 20 to 24 are most likelyto be eligible (68 of all new mothers in this age cate-gory; 80 percent of those in the labour force in 1989).The drop in potential eligibility for older new mothersis probably due to restricted labour force participationassociated with previous children.

New mothers with less than a high school diplomaare much less likely to be eligible for maternity UI(only 40 percent of all new mothers are eligible; 61percent of labour force participants) than are newmothers in any other educational group. Surpris-ingly, new mothers with a university-level educationare relatively unlikely to qualify.

Women reporting that they experience difficultyin obtaining as many weeks of employment asdesired are much less likely than other women to beeligible for UI benefits. Only 37.4 percent of all newmothers reporting themselves “constrained” in termsof weeks of work would be eligible; 31.5 percent ofnew mothers in the labour force reporting themselves“constrained” would be able to satisfy all conditions



TABLE 2Probit Estimates of the Probability of Having a Baby in 1990 – Women aged 16-64

Specification 1 Specification 2

Eligible for maternity benefits under the UI system1 0.045 -0.093(0.068) (0.059)

Replacement rate -0.058 -0.049(0.113) (0.113)

Weeks worked in 1989 (includes self-employment weeks) -0.004*** –(0.001)

Predicted wage, 1989 0.179*** –(0.055)

Average hours worked/week, 1989 -0.002 -0.002(0.002) (0.002)

Age2

16 -0.400** (0.168) -0.473*** (0.162)17-19 -0.152* (0.084) -0.176** (0.079)20-24 0.084* (0.045) 0.099** (0.042)35-44 -0.991*** (0.055) -1.123*** (0.053)45 -64 -2.015*** (0.110) -2.049*** (0.108)

Children 0.256*** 0.100(0.058) (0.107)

No. of children -0.480*** -0.231***(0.036) (0.042)

Children aged 0-2 0.441*** 0.269***(0.047) (0.081)

Children aged 3-5 0.298*** -0.108*(0.052) (0.066)

Not English -0.061* -0.052(0.033) (0.033)

Disabled -0.037** -0.036**(0.016) (0.016)

Minority 0.399*** 0.405***(0.055) (0.055)

Other family income -1.54E-06* -0.000001(8.876E-07) (0.0000008)

Single -0.935*** -0.959***(0.048) (0.049)

Education3

Less than high school 0.151* (0.084) 0.062 (0.081)Some high school 0.112** (0.053) 0.057 (0.052)Some postsecondary -0.037 (0.054) -0.042 (0.054)Certificate/diploma 0.085* (0.047) 0.122*** (0.046)University degree -0.099*(0.059) 0.005 (0.054)Trade -0.116 (0.093) -0.127 (0.092)

... continued



for UI eligibility. This contrasts with 64.6 percentof “non-constrained” new mothers (78.2 percent of“non-constrained” new mothers in the labour force)who would be eligible.

For new mothers with any labour force attach-ment in 1989, some of the largest differences inpotential eligibility for benefits exist across occu-pation and industry groups. For example, 84 percentof new mothers with clerical occupations and 81percent of new mothers with professional occupa-tions are eligible for maternity UI while only 36percent of new mothers with farming or fishing oc-cupations are eligible. (No new mothers with theoccupation of full-time homemaker would be eligi-ble for maternity benefits in Canada.)

For women with some labour force attachmentin 1989, there are also important differences across

industry of employment in the proportion of the sam-ple eligible for maternity/parental benefits. Forexample, 94 percent of women employed in utili-ties were eligible; only 45 percent of womenemployed in primary industries were eligible.

Certain characteristics of the family also appearto have important links with potential eligibility formaternity/parental benefits. First, new mothers whoare not married are less likely than those who aremarried or living with a partner to be eligible formaternity/parental benefits. Second, number of chil-dren present in the family in addition to the newbornis extremely important. Seventy-eight percent offirst-time new mothers are eligible for benefits; 58percent of new mothers with one other child areeligible; 43 percent of new mothers with two otherchildren are eligible; only 11 percent of new motherswith three or more other children are eligible.

TABLE 2(Cont’d.)

Specification 1 Specification 2

Intercept -1.729*** -0.822***(0.308) (0.062)

No. of observations at one 1080 1080No. of observations at zero 25566 25566

Association of predicted probabilities and observed responsesConcordant 85.8% 85.4%Discordant 13.2% 13.6%Tied 1.1% 1.0%

Testing global null hypothesisAIC 7302.333 7404.242SC 7515.283 7600.812-2 Log L 7250.333 7356.242

Notes: Figures in parentheses are (asymptatic) standard errors.1. UI eligibility is based on 1989 work history.2. The excluded variable for age is age 25-34.3. The excluded variable for education is high school diploma.

*Indicates the variable is significant at the 90 percent confidence level.**Indicates the variable is significant at the 95 percent confidence level.

***Indicates the variable is significant at the 99 percent confidence level.



Finally, Figure 1 illustrates a clear relationship be-tween hourly wage rate in 1989 and potentialeligibility for maternity/parental benefits in 1990.Women with low hourly wages are very unlikely tobe eligible while women with higher hourly wagesare much more likely to be eligible.

BEHAVIOURAL RESPONSES TO MATERNITY/PARENTAL BENEFITS

Fertility Responses?Neoclassical economic theory, following Becker(1981), argues that a reduction in the opportunitycost of having children, through, for example, theprovision of maternity benefits, should increase

childbearing. In the Canadian context, two studiesof this issue using aggregate time series data reachdifferent conclusions. Hyatt and Milne (1991) findthat the introduction of maternity benefits in 1971did increase the total fertility rate in Canada, thoughfertility was inelastic with respect to benefit levels.In contrast, Zhang, Quan and Van Meerbergen(1994) find no evidence of such an effect.15

Since possible fertility responses are central toan understanding of the economic consequences ofthe maternity/parental benefits program, we re-visitthis issue using the LMAS micro data. The ques-tion addressed at this stage is: Is a woman morelikely to have a baby if she will receive maternitybenefits? Table 2 presents probit estimates of the

FIGURE 1Percentage of Women with a Baby in 1990 and in the Labour Force in 1989 Eligible for Maternity Benefitsby 1989 Hourly Wage Decile

90

100

80

70

60

50

40

30

20

10

01 2 3 4 5 6 7 8 9 10

Decile

% o

f D

ecile

Po

pu

lati

on

EIUI



probability of having a baby in 1990, using the fullLMAS sample of women aged 16 to 64 (26,646 ob-servations).16 The key explanatory variable is adummy which indicates that the woman is poten-tially eligible for maternity benefits under the formerUI system (which was in place at the time the datawere collected). However, we also examine the ef-fect of the replacement rate (while the specifiedreplacement rate was 55 percent in 1990, a ceilingon benefits means that higher-income women effec-tively receive a lower rate). If incentive effects exist,they should be smaller for women receiving lowerreplacement rates.

Two specifications of the fertility equation arereported in Table 2. In the first, we include weeksworked in 1989 (including self-employment weeksor weeks with low hours or earnings, which do notcount toward benefit eligibility), and wage rate in1989.17 While these variables are used in the calcu-lation of eligibility status, they might also beexpected to exert an independent influence on fer-tility. However, given the strong correlation betweenweeks and wages and eligibility for benefits, we alsoreport a specification that excludes these variables.Other controls include average hours worked perweek in 1989, age, number and age of other chil-dren, marital status, and other family income.

The key point to take from Table 2 is that neitherthe “eligible for benefits” dummy variable nor thereplacement rate are significant in either specifica-t ion. Addit ional weeks of work in 1989 areassociated with lower fertility; a higher hourly wagerate in 1989 is associated with higher fertility.18

Thus, women who worked more weeks in 1989 areless likely to have babies in 1990; women withhigher hourly wages in 1989 are more likely to havebabies in 1990, but there is no statistically signifi-cant difference in estimated fertility between twowomen with the same weeks and/or hourly wage,one of whom is eligible for benefits and one of whomis not (e.g., because one is self-employed or becauseone works many weeks at low hours/wages). Thus,the micro data results support the conclusions of

Zhang, Quan and Van Meerbergen (1994): the avail-ability of maternity benefits does not appear toincrease the probability of having a baby.

This does not seem an unreasonable finding. Thedecision to have a child (to the extent that this isrationally planned) is a major one. While a womanwho claims the full 25 weeks of benefits will re-ceive a substantial amount of money (e.g., $4,592for a woman with the average weekly earnings forwomen with babies in 1990),19 this is still smallcompared with the overall cost of the child, to saynothing of non-pecuniary costs such as sleeplessnights, diaper changes, etc.

Perhaps more reasonable is the finding of a Swed-ish author (Hoem 1990), that the structure of amaternity benefits program is likely to influence thetiming of births for individuals already planning tohave children. This is essentially the question ad-dressed in the next section: Might a woman planningto have a baby in 1990 adjust her labour supply be-haviour in 1989 in order to ensure entitlement tomaternity/parental benefits?

Labour Supply Responses?Little is currently known about possible labour-supply effects of the Canadian maternity/parentalbenefits program. This section asks whether womenmight adjust their labour supply behaviour in orderto gain eligibility for benefits if they are planningto have a baby (assuming having a baby is entirelya planned event)? Under the former UI system, forexample, might a woman who had previouslyworked only ten hours per week in the paid labourmarket, try to increase labour supply in the yearbefore the anticipated birth of a child to the neces-sary 15 hours per week (or to the necessary 20 weeksper year, or both)?

To explore this issue, we first examined the fre-quencies of paid weeks worked per year and paidhours worked per week for all women, and forwomen with babies in 1990. Since the data reflect aperiod during which the UI system was in place, one



FIGURE 2aWeeks Worked in 1989, Excluding Those Who Worked 53 Weeks or Zero Weeks

FIGURE 2bAverage Hours per Week Worked 1989, Excluding Those Who Worked Zero Weeks

6

7

5

4

3

2

1

01

Women With Baby in 1990

All Women

3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51

25

20

15

10

5

01

Women With Baby in 1990

All Women

3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49



might expect to see spikes at 15 hours per week and20 weeks per year for new mothers if significantstrategic use of the maternity benefits program hadgone on. Neither is apparent (see Figures 2a and 2b).

In any case, working any particular combinationof hours and weeks is not necessarily an indicationof strategic behaviour designed to take advantageof the maternity/parental benefits system. A womanwho had, for some reason, chosen to work 15 hoursper week and 20 weeks per year and who had doneso for ten years prior to the birth of her child canhardly be accused of strategic behaviour if she thenclaims benefits. The label of strategic behaviourmight more appropriately be given to someone whoincreased labour supply in the year prior to the birthof a child in such a way as to satisfy the 15-hourand 20-week conditions. Of course, even this mightbe explained by other factors, such as finishing aneducation or getting divorced.

We again use the LMAS micro data to look forevidence that women who were planning to havebabies in 1990 adjusted their labour supply behav-iour between 1988 and 1989 by estimating a probitequation with dependent variable equal to 1 if awoman either increased hours per week to somenumber greater than or equal to 15 and/or increasedweeks per year from less than 20 to 20 or more be-tween 1988 and 1989 so that she would have beenineligible for benefits according to 1988 labour sup-ply but eligible according to 1989 labour supply.These changes can be from zero weeks/hours to over20/15, respectively, or from any small positivenumber of paid weeks/hours. The estimating sam-ple includes all women aged less than 65.

How common are changes of this sort? While 7.4percent of all women (9.1 percent of women whohad babies in 1990) worked less than 20 weeks in1988, but more than 20 weeks in 1989 and 7.2 per-cent of all women (5.8 percent of women who hadbabies in 1990) worked less than 15 hours in 1988but more in 1989, only 2.1 percent of all women(2.3 percent of women who had babies in 1990) in-

creased both hours and weeks. That is, only 2.3 per-cent of women who had babies in 1990 would nothave been eligible for benefits based on 1988 be-haviour but were eligible based on 1989 behaviour.No obvious pattern of women who are planning tohave babies modifying their behaviour to a greaterdegree than other women is immediately obvious.The key explanatory variable included in the probitequation for increasing labour supply to “becomeeligible” for maternity/parental benefits is a dummyvariable indicating that the woman actually had ababy in 1990: our proxy for having planned to havea baby. It is, of course, bizarre to suppose that allpregnancies are rationally planned well in advance.Many “accidents” happen. Moreover, many couplesplanning to have children are unable to do so, cer-tainly at a very particular time. But, if we are toinvestigate the possibility that some individuals whoare planning to have babies adjust their behaviourso as to take advantage of the maternity/parentalbenefits system, then we have to assume a fair ca-pacity to plan for pregnancy.

Of course, other factors than a hoped-for preg-nancy may lead to increases in labour supplybetween one year and the next. Thus, we include asexplanatory variables indicators of change in edu-cational status: (i) dropped out indicates that theindividual attended school full-time in 1988, but notin 1989 with no change in educational credentials;(ii) graduated indicates that the individual attendedschool full-time in 1988, but not in 1989 and in-creased educational credentials (e.g., moved fromsome postsecondary to university degree); (iii) cor-respondence/part-time indicates that the individualdid not attend school full-time in either year, butnevertheless increased educational credentials. (Thebase category is for those with no change in school-ing, either because they were not at school in eitheryear or because they were at school in both years,with no change in credentials.)

We also include two indicators of change in mari-tal status: became married between 1988 and 1989and became divorced between 1988 and 1989, since



either change in marital status may lead to changesin labour supply behaviour. Interprovincial migra-tion may also be associated with increased laboursupply if, for example, the woman had moved totake up a job. Finally, changes in perceived labourmarket constraints, reflecting changes in local la-bour demand conditions may be associated withincreased labour supply from year to year.

Results of the probit analysis are presented in theAppendix, Table A2. The first set of regressionsstraightforwardly include the dummy variable indi-cating that the woman had a baby in 1990. However,it is clear from the probit results reported in TableA2 that fertility behaviour depends upon labour sup-ply. In an effort to deal with this endogeneity, thechange in labour supply equation was also estimatedreplacing the actual zero/one 1990 fertility dummyvariable with the predicted probability of having ababy as calculated using a first-stage fertility equa-tion available on request from the author.20

The effects of control variables are unremarkable(increases in labour supply are associated with in-creases in education, divorce, and, especially,removal of labour market constraints).21 Becomingmarried, or moving interprovincially are not statis-tically significant. A dummy variable indicating thatthe woman had a baby in 1990 is statistically insig-nificant whether entered directly or through aninstrumental variables procedure. Thus, it appearsthat women planning to have a baby in 1990 did notincrease labour supply between 1988 and 1989 inorder to take advantage of the maternity/parentalbenefits program. Given the very high rates of la-bour force participation by young women, and evenby mothers with young children, this result presum-ably reflects the fact that there are not that manyyoung women outside the labour force who wouldbe able to enter in order to take advantage of thematernity benefits program.22

CHANGES IN POTENTIAL ACCESS TO BENEFITS

RESULTING FROM EI

Figure 3 illustrates the difference between UI andEI in terms of eligibility conditions for maternity/parental benefits. Under UI, claimants required 20weeks of work and 15 hours per week.23 Thus, inFigure 3, anyone located in the upper-right quad-rant of the diagram would be eligible for benefits.Under EI, claimants require 700 hours of employ-ment. The 700-hour rectangular hyperbola is notedin the diagram; anyone located to the right of thiscurve is eligible for maternity/parental benefits un-der the EI system. Assuming no change in labour-supply behaviour, claimants whose entitlement sta-tus would change as a result of the switch from UIto EI are those located in any of the shaded areas.First, workers with fewer than 15 hours per week,but with enough weeks to have accumulated 700hours would qualify under EI but not under UI (seethe shaded area on the right of the diagram abovethe rectangular hyperbola but beneath the horizon-tal line).24 Note, however, that a woman working14 hours per week would require 50 weeks of paidemployment to establish entitlement. A secondgroup of workers will also benefit from the switchfrom UI to EI: those with less than 20 weeks of eli-gible employment but sufficient hours per week tototal 700 (see the shaded area on the left side of thediagram above the rectangular hyperbola but to theleft of the vertical line). For example, a womanworking only 18 weeks, but 40 hours per week wouldbe eligible for maternity/parental benefits under EIbut not under UI. On the other hand, disentitlementwill occur for women working at least 20 weeks withat least 15 hours per week but without the 700 hoursrequired for EI (see the shaded area underneath therectangular hyperbola but above both the 15 hourand 20 week lines). For example, a new mother withexactly 15 hours and 20 weeks of employment willhave substantially less (300 hours) than the 700minimum hours required for benefits under the EIsystem.



Given the results of the previous section, it seemsunlikely that any major changes in either fertility orlabour supply would have followed the programchange. Thus, Table 3 once again uses the 1989/90LMAS data simply to calculate, based on 1989 em-ployment history, whether women who had babiesin 1990 would have been eligible for maternity/parental benefits under the UI system and under theEI system. It is important again to note that we arenot using data which span the actual time of thepolicy implementation. Rather, given the economet-ric evidence that it is unlikely that major behaviouraladjustments will have occurred as a result of thechange in the maternity/parental benefits program,this exercise asks how women living with the la-bour market conditions of 1990 would have been

affected had they been faced with an EI rather thana UI system.25

Results suggest relatively little difference (1.6 per-centage points) in potential eligibility for maternity/parental, for new mothers in the labour force in theyear prior to the birth of their child. In terms of Figure3, 1.8 percent of women became entitled to benefitsbecause while they did not have 20 weeks of employ-ment, they worked enough hours per week to reach700 (i.e., they were located in the left-most lightlyshaded area). Only 0.6 percent of the sample gainedeligibility because they had many weeks, but less than15 hours per week. On the other hand, 4 percent lostentitlement as a result of the switch from UI to EI (werelocated in the darker shaded area of Figure 1).

FIGURE 3Eligibility Requirements – Comparing the UI and the EI Systems

70

60

50

40

30

20

10

02 4 6 8

Weeks/Year

Wee

kly

Ho

urs

0 12 14 16 1810 22 24 26 2820 32 34 36 3830 42 44 46 4840 50 52

Individuals who qualifiedunder UI (20 weeks, 15

hours/wk) but do notqualify under EI (700

Hours/wk)

Individuals who qualifyunder EI (700hours/year)

who did not qualifyunder UI (20 weeks,

15 hours/wk)

Individuals who qualifyunder both the EI and UI systems

700 Hours/Years



TABLE 3Women with a Baby in 1990 Potentially Qualifying for Maternity and/or Parental Benefits Based on 1989 EmploymentHistory – UI versus EI (Weighted)

Women in the Labour Force in 1989 with a Baby in 1990

Eligible Based on UI Maternity/Paternity Eligible Based on EI Maternity/PaternityBenefit Requirements Benefit Requirements (700 Hours)

All women with a baby in 1990and 1989 in labour force 73.3 71.7

Age16-19 years 46.3 44.120-24 80.2 76.825-34 72.4 72.035-64 75.7 71.5

EducationLess than high school diploma 61.1 62.8High school d iploma 76.9 73.6Some postsecondary 78.5 77.1Certificate/trade/diploma 75.0 75.1University degree 70.8 65.2

ProvinceAtlantic 71.5 68.9Quebec 81.1 80.3Ontario 68.7 70.0Manitoba 75.1 68.6Saskatchewan 73.5 72.8Alberta 74.4 66.8British Columbia 69.7 64.5

Underemployed in 1989 31.5 34.1Not underemployed in 1989 78.2 76.1Occupation

Management/administrative 77.0 86.1Professional 81.0 74.7Clerical 83.5 82.5Sales/service 69.8 64.8Other occupation 62.3 65.1Farm/fisheries 36.0 31.9

IndustryPrimary 45.3 43.1Manufacturing 75.5 77.8Utilities 94.0 88.1Retail/wholesale trade 74.9 73.5Finance 82.0 86.4Social services 77.9 72.6Other services 70.2 67.4Government 87.1 83.8

Number of Children0 79.9 80.41 68.6 65.32 66.1 61.93+ 30.7 22.5

Marital StatusMarried 73.6 72.1Not married 70.5 68.2



Therefore, it is unlikely that the switch from UIto EI has resulted in major changes in overall ac-cess to maternity/parental benefits. There are,however, some interesting variations in the conse-quences of the program change for women inparticular circumstances. For example, while theprogram change has almost no effect on new moth-ers aged 25 to 34, fewer younger (less than 25) andolder women (more than 35) will have access tomaternity/parental benefits (reductions of 3.4 and4.2 percentage points respectively). Women report-ing themselves constrained in terms of weeks ofemployment in 1989 are more likely (2.6 percent-age points) to receive EI than UI. There is aneight-percentage point reduction in potential eligi-bility for new mothers with three or more otherchildren; a four-percentage point reduction for thosewith two other children.

To better sort out these findings, Table 4 presentsthe results of probit analyses of the probability ofqualifying for maternity/parental benefits under thetwo policy regimes. In column 1, the dependent vari-able is a dummy equal to one if the woman wouldhave qualified for UI maternity/parental benefits in1990, based on 1989 employment history; in col-umn 2, the dependent variable is a dummy equal toone if the woman would have qualified for EImaternity/parental benefits based on 1989 employ-ment history. The sample for analysis is all womenwho gave birth to (or adopted) a baby in 1990 andwho had some labour force experience in 1989.

A comparison of the probit equation for UI eli-gibility with the probit equation for EI eligibilitysuggests an overall similarity in the characteristicsassociated with a higher (or lower) probability ofbeing eligible for benefits, with a few notable dif-ferences. In both cases, teenaged new mothers aremuch less likely to be eligible for benefits. Newmothers without a high school education are lesslikely than those with a high school education to beeligible for maternity benefits, though this effect islarger under the EI system. Rather unexpectedly,new mothers with university-level education are also

less likely to be eligible than new mothers with ahigh school diploma. Perhaps women with a uni-versity degree have had less opportunity to work inthe paid labour force before having children? Witheither program, being unable to obtain as manyweeks of paid employment as desired has a largenegative association with being eligible for benefits.And, having any other children is negatively relatedto the probability of being eligible for benefits; ad-ditional other children increase the magnitude of thiseffect. In general, women who are least well-off arealso least likely to be eligible for maternity/parentalbenefits. 26

CONCLUSIONS

In Canada, maternity and parental benefits are pro-vided through the Employment Insurance program(previously Unemployment Insurance), though eli-gibil i ty conditions are generally tougher formaternity or parental benefits than for regular EI.This feature of the program design reflects an his-torical interest in minimizing incentives for womento enter the labour force only for the purpose of gain-ing access to benefits. Yet, do we need to worry aboutsuch strategic behaviour? Very little is known aboutthe behavioural implications of the Canadianmaternity/parental benefits system. If, in fact, be-havioural effects are minimal, then it might be areasonable policy option to make it at least as easyto qualify for maternity/parental benefits as toqualify for regular EI benefits, particularly in viewof the relatively bad economic conditions currentlyfaced by Canadian adults aged less than 35, whoare most likely to have babies.

This paper uses micro data from the 1988/89/90Labour Market Activities Survey to study some be-havioural implications of the maternity/parentalbenefits system. We find, first, that fertility behav-iour is not significantly influenced by the availabilityof maternity/parental benefits and, second, that thereis no evidence that women adjusted their laboursupply behaviour in order to gain access to benefits



TABLE 4Probit Estimates of the Probability of Satisfying Qualifying Conditions – All Women in the Labour Force with a Baby in 1990Sample = 884

Variable Qualifying for Unemployment Insurance* Qualifying for EI

Intercept 0.682** 0.725***(0.268) (0.264)

Provincial unemployment rate – females 0.026 -0.0003(0.023) (0.022)

Age1

16-19 -0.894*** (0.242) -0.940*** (0.240)20-24 0.165 (0.127) 0.009 (0.126)35-64 0.225 (0.222) 0.148 (0.217)

Education2

Less than high school -0.294 * (0.164) -0.021 (0.164)Some postsecondary -0.007 (0.178) 0.063 (0.177)Certificate/diploma/trade -0.241* (0.145) -0.102 (0.144)University degree -0.492*** (0.188) -0.488*** (0.187)

Occupation3

Managerial/administrative -0.041 (0.180) 0.411** (0.199)Professional 0.394** (0.175) 0.268 (0.171)Sales/services -0.118 (0.163) -0.209 (0.162)Farming/fishing -0.257 (0.446) -0.241 (0.444)Other occupation -0.783* (0.255) -0.648** (0.257)

Industry4

Primary -0.296 (0.271) -0.228 (0.269)Manufacturing 0.798*** (0.265) 0.797*** (0.268)Retail/wholesale trade 0.282 (0.177) 0.293* (0.175)Utilities 1.095*** (0.382) 0.813** (0.329)Finance 0.172 (0.181) 0.408** (0.187)Other services 0.089 (0.198) 0.163 (0.198)Government 0.551** (0.264) 0.497** (0.252)

Other family income 2.821E-06 3.369E-06(2.512E-06) (2.511E-06)

Underemployed -1.154*** -0.985***(0.163) (0.162)

Additional children5

One -0.392*** (0.116) -0.523*** (0.115)Two -0.356** (0.163) -0.557*** (0.160)Three or more -1.207*** (0.316) -1.491*** (0.333)

Visible minority 0.123 0.021(0.211) (0.208)

Immigrant -0.009 0.172(0.180) (0.182)

Disabled -0.175 -0.164(0.256) (0.255)

... continued



TABLE 4(Cont’d.)

Variable Qualifying for Unemployment Insurance* Qualifying for EI

French 0.289** 0.562***(0.138) (0.139)

Other language -0.249 -0.291(0.184) (0.185)

Single 0.433** 0.411**(0.203) (0.199)

In school full-time in 1989 -0.268 -0.301(0.228) (0.224)

Dependent variable(total weight)=1 occurred 73.3% 71.7%=0 did not occur 26.7% 28.3%

Association of predicted probabilities andobserved responses

Concordant 76.5% 76.5%Discordant 23.3% 23.3%Tied 0.3% 0.2%

Notes: *In column 1, the dependent variable is a dummy=1 if the woman would have qualified for maternity/parentalbenefits under the UI system in 1990, based on her employment history in 1989. In column 2, the dependentvariable is a dummy=1 if the woman would have qualified for EI, based on 1989 employment history. Theestimating sample is all women who had a baby in 1990 and who were in the labour force at some time in 1989.1. The excluded variable for age is 25-34.2. The excluded variable for education is high-school diploma.3. The excluded variable for occupation is clerical.4. The excluded variable for industry is social services.5. The excluded variable for additional children is none.

Standard Errors in parentheses*indicates variable is significant at 90% confidence level.

**indicates variable is significant at 95% confidence level.***indicates variable is significant at 99% confidence level.

under the former UI system (which was in opera-tion at the time the data used for the analysis werecollected).

We also examine who is potentially eligible formaternity/parental benefits and find that teenagednew mothers and those with little education are muchless likely to have access to benefits. Women who

are experiencing difficulty in the labour market inthe sense of being unable to obtain as many weeksof paid employment as desired are less likely to beeligible for benefits. Women with other children areless likely to be eligible; this effect becomes largeras the number of other children increases. Thesebasic conclusions are true for the former UI systemas well as for the EI system. (There appears to be



little difference between the programs in terms ofaccess to benefits.) In light of the evidence againstlarge behavioural responses, it thus seems reason-able to ease access to maternity/parental benefits.

NOTES

I would like to thank Lynn Lethbridge and Janice Yatesfor excellent research assistance. Peter Burton and par-ticipants at seminars at Brock University, McMasterUniversity, and Dalhousie University provided very help-ful comments. The Social Sciences and HumanitiesResearch Council of Canada as well as Human ResourcesDevelopment Canada provided the financial support nec-essary for completion of this research.

1Notice that while higher regional unemployment ratesmake it easier to receive regular UI benefits, this is nottrue for maternity benefits.

2See Routhier and Labowka (1994) for further details.These authors point out that the 1984 decision to elimi-nate the magic-10 rule was in part the result of publicpressure following the Supreme Court ruling in the caseof Bliss v. the Attorney General of Canada that “discrimi-nation on the basis of pregnancy is not discrimination onthe basis of sex.” Bliss had argued that the 20-week workrequirement was discrimination on the basis of sex.

3The economic consequences of maternity and paren-tal cash benefits programs have, in general, receivedrelatively little attention, though there has been some re-cent interest in parental leave programs. See, for example,Dalto (1989); Gruber (1994); and Waldfogel (1997) whostudy implications of parental leaves for women’s wages/earnings; Garrett, Wenk and Lubeck (1990) and Klermanand Leibowitz (1994 and 1997) who study implicationsof parental leaves for labour force participation aroundthe time of childbirth. Ruhm and Teague (1997) and Ruhm(1998) use international comparisons to study some ofthe broader economic consequences (e.g., economic effi-ciency; women’s employment and relative wages) ofvariations in maternity/parental leaves and benefits.

4See Hyatt and Milne (1991) and Zhang, Quan andVan Meerbergen (1994), discussed later in the paper.

5Perhaps because maternity and parental benefits arepaid through the EI/UI system, there is seldom as much

attention paid to goals for this sub-program. While regu-lar benefit payments do constitute the larger budgetarycomponent, maternity/parental benefits averaged from 10to 15 percent of total benefits paid in any month of 1996(CANSIM, D730311,D736408, D739804, D730283).

6Empirically, we cannot distinguish births from adop-tions, which should not matter since parents who adoptare eligible for parental benefits on the same terms.

7Alternative data sources would be the Survey of La-bour and Income Dynamics (SLID) and the COEP.Although these data sets are newer, public-use versionsdo not have enough information for the purposes of ouranalysis. The COEP, moreover, does not include any newmothers who have not recently been in the labour marketwhich would preclude the fertility analysis conductedhere. We do not yet have public access to enough yearsof SLID data to conduct this analysis.

8Phipps (1994) reports that men were more likely thanwomen to be eligible for parental benefits, though veryfew actually take these benefits.

9For example, it is not possible to distinguish the caseof a woman marrying and giving birth within two yearsand the case of a woman marrying a man with a 0-2 year-old child. We considered eliminating “unlikely” births,but decided that this introduced too much researcher bias.

10Routhier and Labowka (1994) is an HRDC evalua-tion document that provides a careful history and raisesmany policy questions. Moloney (1989a and 1989b) pro-vides some descriptive evidence about maternity absencesusing the Absence from Work Survey. Pal (1985) argues,from an historical and analytical perspective, that mater-nity benefits should not be included as part of the UIprogram since UI is governed by issues of labour forceattachment while maternity benefits are to aid familiesaround the time of the birth of a child.

11While parental benefits were not actually availableuntil November 1990, it is still possible to ask how manywomen could potentially have been entitled, given theeligibility conditions eventually established.

12Although Canadian maternity/parental benefits op-erate as a labour market program, this analysis considerspotential eligibility for a sample of all women with in-fants because discouraged worker phenomena in ahigh-unemployment region, for example, may mean that



fewer individuals enter the labour force. If we simplyexclude these individuals from the denominator, we willover-estimate true access to maternity/parental benefitsin the region. Moreover, it is important to remember thatin some countries, maternity/parental benefits are avail-able to all new parents: not just those with sufficientlabour market attachment.

13Since only the labour market data are available on aweekly basis, this choice seemed the cleanest for two rea-sons: First, we can then follow the same procedure for allwomen. Second, we can discuss the same time period forall variables.

14It may well be the case that women who are eligiblefor maternity/parental benefits are more likely to take timeout of the labour market. For new mothers who had beenin the labour force in 1989, potential eligibility is higherfor those who took time out than for those who did not(79.4 percent versus 66.3 percent).

15Hyatt and Milne use a variable constructed as aweighted average of average female wages and averagematernity benefits, post-1970. Zhang, Quan and VanMeerbergen experiment with dummy variables for intro-duction/liberalization of the program.

16The estimating sample includes all women aged lessthan 65 since there was a small number of women in the45-64 age ranges reporting the addition of an infant tothe household. While giving birth is less likely for theolder women, adoption is a possibility that would entitlethe new mother to benefits. These equations were alsorun using a sample of women aged less than 45, and re-sults were essentially unchanged.

17Since not all women in the sample are observed withpositive wages, a predicted wage rate for 1989 is used.The wage equation is estimated using the sample of allwomen in the LMAS with positive wages in 1989, cor-recting for sample selectivity bias in the usual waythrough the inclusion of an inverse mill’s ratio from afirst-stage, labour force participation probit analysis.Explanatory variables in the wage equation include: dum-mies for age, region, education, occupation, and industryas well as number of other children. The first-stage probitfor labour force participation in 1989 includes weeksworked in 1988, disability status in 1989, minority statusand marital status as well as the age and education dum-mies. These results are available upon request from theauthor.

18If we take out all of the labour market variables, theUI eligibility dummy becomes statistically significant, butnegative: presumably because weeks of work is statisti-cally significant and negative and UI eligibility dependsvery much on past weeks of work.

19This is calculated as 55 percent of $334 for 25 weeks.

20Instruments include woman’s age, number and ageof other children, minority status, language, and disabil-ity status.

21Labour supply is also predicted to increase when alabour market constraint has appeared rather than disap-peared. This seemingly counter-intuitive result in factmakes sense because it means that the individual is nowlooking for more work (and may previously not havewanted paid employment at all). Thus, for example, 19percent of women who were underemployed in 1989, butnot in 1988, had more weeks of paid employment in 1989.

22In 1989, 94.5 percent of women aged 16 to 64 with-out children were in the labour force; 75.9 percent ofwomen aged 16 to 64 with children were in the labourforce.

23It is assumed that the wage rate is exogenous in therelatively short run, so no attention is paid to the possi-bility of an individual attempting to achieve a higher wagerate in order to satisfy the minimum earnings conditionof UI.

24Unless they had very high earnings.

25Unemployment for women was lower in 1999 thanin 1990 (7.3 versus 8.1 percent — CANSIM D980759).Female labour force participation was very similar (58.9in 1999 versus 58.5 in 1990 — CANSIM D980792).

26Of course, this does not necessarily mean that theydo have access to any cash transfers. For example, 16.2percent of all women we estimate to be ineligible formaternity benefits report receipt of social assistancebenefits.

REFERENCES

Becker, G.S. 1981. A Treatise on the Family. Cambridge:Harvard University Press.

Dalto, G.C. 1989. “A Structural Approach to Women’sHometime and Experience-Earnings Profi les:



Maternity Leave and Public Policy” Population Re-search and Policy Review 8:247-66.

Garret, P., D. Wenk and S. Lubeck. 1990. “WorkingAround Childbirth: Comparative and Empirical Per-spectives on Parental-Leave Policy,” Child Welfare69(5):401-13.

Gruber, J. 1994. “The Incidence of Mandated Maternity Ben-efits,” The American Economic Review 84(3):622-41.

Hoem, J.M. 1990. “Social Policy and Recent FertilityChange in Sweden,” Population and Development Re-view16(4):735-48.

Hyatt, D.E. and W.J. Milne. 1991. “Can Public PolicyAffect Fertility?” Canadian Public Policy/Analyse dePolitiques 17(1):77-85.

Klerman, J.A. and A. Leibowitz. 1994. “The Work-Em-ployment Distinction among New Mothers,” TheJournal of Human Resources 29(2):277-03.

______ 1997. “Labour Supply Effects of State MaternityLeave Legislation,” in Gender and Family Issues inthe Workplace, ed. F.D. Blau and R.G. Ehrenberg. NewYork: Russell Sage Foundation, pp. 65-85.

Moloney, J. 1989a. “Maternity,” Canadian Social Trends(Autumn):30-34.

______ 1989b. “On Maternity Leave,” Perspectives onLabour and Income 1:27-46.

Pal, L.A. 1985. “Maternity Benefits and UnemploymentInsurance: A question of Policy Design,” CanadianPublic Policy/Analyse de Politiques 11(3):551-60.

Phipps, S. 1994. “Potential Access to Maternity and Pa-rental Benefits.” Report prepared for Human ResourcesDevelopment Canada.

Routhier, A. and S. Labowka. 1994. “Unemployment In-surance Provision of Special Benefits: EvaluationSynthesis and Issues,” Ottawa: Insurance ProgramsDirectorate Evaluation Branch, Strategic Policy, Hu-man Resources Development Canada.

Ruhm, C.J. 1998. “The Economic Consequences of Pa-rental Leave Mandates: Lessons from Europe,” TheQuarterly Journal of Economics 113(1):285-317.

Ruhm, C.J. and J.L. Teague. 1997. “Parental Leave Poli-cies in Europe and North America,” in Gender andFamily Issues in the Workplace, ed. F. D. Blau andR.G. Ehrenberg. New York: Russell Sage Foundation.

Waldfogel, J. 1997. “Working Mothers Then and Now: ACross-Cohort Analysis of the Effects of MaternityLeave on Women’s Pay,” in Gender and Family Issuesin the Workplace, ed. F.D. Blau and R.G. Ehrenberg.New York: Russell Sage Foundation.

Zhang, J., J. Quan and P. Van Meerbergen. 1994. “TheEffect of Tax-Transfer Policies on Fertility in Canada, 1921-88,” The Journal of Human Resources 29(1):181-201.



APPENDIX

TABLE A1Sample Characteristics Labour Market Activities Survey, 1990 (Weighted)

All Women Aged Women with a Baby Women in the Labour Force16-64 in 1990 in 1990 in 1989 with a Baby in 19901

Age16-19 8.5 5.2 5.520-24 11.7 27.2 27.725-34 27.2 60.9 60.835-64 52.6 6.6 5.9

EducationLess than high school 28.2 18.1 14.6High school diploma 25.1 27.5 26.9Some postsecondary 12.7 12.7 12.6Certificate/diploma/trade 21.4 28.0 31.1University degree 12.6 13.7 14.9

ProvinceAtlantic 8.7 7.2 7.4Quebec 26.4 28.6 26.5Ontario 36.9 35.8 37.5Manitoba 3.8 4.0 3.9Saskatchewan 3.4 3.1 3.2Alberta 9.2 9.3 9.7British Columbia 11.4 12.0 11.7

OccupationManagement/administrative 7.6 9.1 11.2Professional 15.6 19.3 23.7Clerical 21.9 23.9 29.5Sales/service 20.8 16.6 20.5Other 6.6 7.7 9.5Farming/fishing 1.7 1.4 1.8No paid occupation: homemaker 25.9 22.0 3.8

IndustryPrimary 3.7 3.9 4.7Manufacturing 7.9 10.2 12.5Utilities 3.2 3.2 3.9Retail/wholesale trade 13.4 13.3 16.4Finance 9.7 11.7 14.4Social services 18.7 21.0 25.9Other services 12.6 10.5 13.0Government 4.8 4.2 5.2No industry: homemaker 25.9 22.0 3.8

Number of Children0 62.2 45.3 51.81 15.7 34.1 33.92 15.8 15.2 12.03+ 6.3 5.4 2.3

Marital statusMarried 67.1 88.2 89.2Not married 32.9 11.8 10.8

Number of observations 26,646 1,080 884

Notes: 1. Labour force participants are those whose weeks of employment plus weeks of unemployment in 1989 aregreater than zero.



TABLE A2Probit Estimates of the Probability of Increasing Labour Supply Behaviour Between 1988 and 1989 to Become Eligiblefor Maternity/Parental Benefits – Women Aged 16-64

Actual Fertility Predicted Fertility

Had a baby in 1990 0.031 –(0.084)

Predicted probability of having a baby in 1990* – 0.198(0.266)

Dropped out 0.251*** 0.249***(0.087) (0.087)

Graduated 0.360** 0.359**(0.163) (0.163)

Correspondence/part-time -0.076 -0.077(0.059) (0.059)

Became married between 1988 and 1989 surveys 0.145 0.132(0.107) (0.109)

Became divorced between 1988 and 1989 surveys 0.411*** 0.416***(0.117) (0.117)

Underemployed in 1989, but not in 1988 0.240*** 0.239***(0.065) (0.065)

Underemployed in 1988, but not in 1989 0.504*** 0.504***(0.049) (0.049)

Migrated interprovincially in 1989 -0.218 -0.222(0.226) (0.226)

Intercept -2.143*** -2.150***(0.023) (0.025)

No. of observations at one 560 560

No. of observations at zero 26086 26086

Association of predicted probabilities and observed responsesConcordant 41.2% 49.9%Discordant 20.6% 27.1%Tied 38.2% 23.1%

Testing global null hypothesisAIC 5185.140 5184.726SC 5267.044 5266.630-2 Log L 5165.140 5164.726

Notes: (Asymptatic) standard errors are in parentheses.1. The dependent variable is a dummy=1 if either hours of paid work increased from less than 15 to more than

15 and/or weeks per year increased from less than 20 to more than 20 between 1988 and 1989 so that awoman previously not eligible for maternity/parental benefits would become eligible.

*Indicates the variable is significant at the 90 percent confidence level.**Indicates the variable is significant at the 95 percent confidence level.

***Indicates the variable is significant at the 99 percent confidence level.

maternity and parental benefits in canada: are there...

Documents