Helena Skyt Nielsen, PhD Professor A A R H U S U N I V E R S I T Y S c h o o l o f E c o n o m i c s a n d M a n a g e m e n t Causes and consequences.

A A R H U S U N I V E R S I T Y

S c h o o l o f E c o n o m i c s a n d M a n a g e m e n t

Helena Skyt Nielsen, PhDProfessor

Causes and consequences of fathers’ child leave: Evidence

from a reform of leave schemes

Spring 2009Very preliminary

Helena Skyt Nielsen, Aarhus University




Average child leave in 1994-2005

0

50

100

150

200

250

300

350

1994 1996 1998 2000 2002 2004

Year

Days

Fathers Mothers




Average share of total leave in 1994-2005

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

Year

Average share

Fathers Mothers




Outline

Previous studies The reform Identification strategy Data Empirical analysis Conclusion




Previous studies

Individual eligibility to leave increases fathers’ usage of child leave Sweden - Ekberg, Eriksson and Friebel (2005) US - Han, Ruhm and Waldfogel (2009)

Father’s leave hurts his own career Sweden - Albrecht et al. (1999), Stafford and Sundström (1996),

Bygren and Duvander (2006) This evidence is mainly suggestive

Father’s leave improves his partner’s career Sweden - Ekberg, Eriksson and Friebel (2005) Sweden - Pylkkänen and Smith (2003) This evidence concerns short-run outcomes




Previous studies

Fathers’ usage of parental leave may influence his children US - Nepomnyaschy and Waldfogel (2007)

Positive effect on short run outcomes: feeding, dressing, bathing, getting up at night

Causal effect? Sweden - Ekberg, Eriksson and Friebel (2005)

No effect on long run outcomes: care of sick children

More generally, fathers’ usage of parental leave may influcence

Gender equality, gender wage gap, workplace organization, etc.




This study

Research questions: Does economic incentives induce fathers to use

parental leave schemes? How much extra monetary compensation is needed to

induce the couple to transfer one day of leave from the mother to the father?

Does a father’s usage of parental leave influence his own career and his wife’s career?




The reform

New legislation - ”flexible child leave scheme” Implemented for children born after March 27, 2002

Content It extended the period with full benefit compensation

From 4+14 + 2+2 +10 weeks To 4+14 + 2+ 32 weeks

It introduced flexibility in the following ways Leave may be shared flexibly between partners and over time Leave may be used simultaneously by both parents




The reform

New interpretation of collective agreements as a consequence of the reform

The mother has a right to wage compensation 4+14 +10

The father has a right to wage compensation 2+2 +10

Conditional on the fact that the employer receives a reimbursement

Reimbursement is now available for 4+14+2+32 weeks (rather than 4+14+2+2+10 weeks)




The reform

I focus on the public sector Two collective agreements Local vs. State Maximum available favorable terms and conditions 100% uncapped wage compensation for a long period of time

Consequence of the reform Parents covered by different collective agreements obtain an

extra 10 weeks of 100% wage compensation if they both use 10 weeks of parental leave

Illustration…




Illustration of the child leave regulations before the reform

MotherFather




Illustration of the child leave regulations after the reform – spouses on same wage agreement

Father

Mother




Illustration of the child leave regulations after the reform – spouses on different wage agreements

Mother Father




Identification strategy

Estimation of the causal effect of economic incentives on leave taking Diff-in-diff

Estimation of the causal effect of father’s leave on career outcome IV

iii uXDifferentPostDifferentPostLeave *210

cellseducationcouplestheforIndicators

rentPost*DiffeDifferentPost

rentPost*Diffe

Leave

vXLeaveOutcome

i

iiii

' )3(

,, (2)

(1)

by edinstrument is




Identification strategy

Identifying assumption Common trend assumption Time-varying variables: preference for leave, business cycle

effects, etc.

Issues to consider Selection into treatment The scheme is complicated – Olsen (2007) states that parents,

employers and local authorities lacked information Delayed response The effect of the ”intention to treat”




Data

Population of Danish individuals 1980-2006 Combined with complete fertility and leave

histories Select all births in 1997-2005 where…

… the mother and father can be identified and form a couple … the mother and father are aged 20 to 50 years … the mother and father completed higher education (short,

medium or long cycle higher education) … the mother and father are publicly employed




Table 1. Sample selection.

Frequency PercentAll live births 1997-2005 614,812 1.00Mother and father can be identified 593,106 0.96Mother and father form couple 543,093 0.88Mother and father are aged 20 to 50 years 534,869 0.87Mother and father have completed higher education 103,961 0.17Mother and father are employed in public sector 18,594 0.03Mother or father is recorded to take leave 18,484 0.03Child birth did not take place Jan 1- March26, 2002 18,077 0.03







Table 2. Descriptive statistics.

Mean Std.dev. Mean Std.dev. Mean Std.dev. Mean Std.dev.Background characteristicsBig City 0.463 - 0.525 - 0.463 - 0.525 -Local 0.805 - 0.209 - 0.805 - 0.791 -Labor income-Benefit per day - - 508.653 346.747 - - 279.953 263.719Other variablesIncome variablesCareer variablesLeave variablesNo leave 0.153 - 0.220 - 0.011 - 0.015 -No parental leave 0.571 - 0.635 - 0.032 - 0.037 -Total leave (days) 28.116 35.330 24.785 33.375 280.132 89.623 280.516 94.464Parental leave (days) 16.474 33.836 14.068 31.432 162.852 90.602 163.795 93.798Share of leave (%) 0.098 0.136 0.093 0.148 0.902 0.136 0.907 0.148

Same agreementFather Mother

Different Same agreement Different

similarsimilarsimilar




Table 2. Continued.

Mean Std.dev. Mean Std.dev. Mean Std.dev. Mean Std.dev.Education variablesSCHE 0.070 - 0.177 - 0.045 - 0.028 -1: SCHE SocSci 0.012 - 0.021 - 0.009 - 0.007 -2: SCHE NatTech 0.021 - 0.040 - 0.012 - 0.018 -3: SCHE PoliceArmy 0.037 - 0.116 - 0.024 - 0.003 -MCHE 0.547 - 0.237 - 0.671 - 0.590 -4: MCHE PedTeach 0.421 - 0.100 - 0.416 - 0.272 -5: MCHE SocSci 0.028 - 0.029 - 0.032 - 0.038 -6: MCHE NatTech 0.038 - 0.095 - 0.012 - 0.015 -7: MCHE Health 0.060 - 0.013 - 0.212 - 0.265 -LCHE 0.383 - 0.585 - 0.284 - 0.383 -8: LCHE Health 0.107 - 0.063 - 0.070 - 0.088 -9: LCHE SocSci 0.111 - 0.200 - 0.090 - 0.117 -10: LCHE NatTech 0.103 - 0.219 - 0.059 - 0.070 -11: LCHE HumSci 0.062 - 0.103 - 0.064 - 0.108 -Number of observations

Father MotherSame agreement Different Same agreement Different

12987 5090 12987 5090




Figure 5. Trends in child leave.

Fathers

0

5

10

15

20

25

30

35

-5 -4 -3 -2 -1 0 1 2 3

Year

Days

Same StateLocal

Mothers

250

260

270

280

290

300

-5 -4 -3 -2 -1 0 1 2 3

Year

Days

Same StateLocal




Table 3. Transition matrices.

A: Pre-reform (before birth) C: Post-reform (before birth)

Period t-1Couple is covered by Same Different Other All Nobs Same Different Other All NobsSame 0.923 0.029 0.048 1.000 6942 0.947 0.026 0.027 1.000 7287Different 0.071 0.867 0.062 1.000 2883 0.110 0.857 0.033 1.000 2735Other combination 0.438 0.207 0.355 1.000 3761 0.426 0.219 0.354 1.000 3724All 0.608 0.256 0.136 1.000 13586 0.639 0.244 0.117 1.000 13746

B: Pre-reform (after birth) D: Post-reform (after birth)

Period t-1Couple is covered by Same Different Other All Nobs Same Different Other All NobsSame 0.807 0.026 0.168 1.000 9143 0.822 0.025 0.153 1.000 8307Different 0.049 0.749 0.202 1.000 3824 0.159 0.634 0.207 1.000 3126Other combination 0.272 0.125 0.603 1.000 2050 0.292 0.121 0.587 1.000 1005All 0.541 0.223 0.236 1.000 15017 0.612 0.186 0.202 1.000 12438

Period tCouple is covered by







Empirical Analysis

The effect of economic incentives on father’s leave taking on mother’s leave takingThe effect of father’s leave usage on mother’s career on father’s career




Table 4. Estimation of the effect of economic incentives on fathers’ leave taking

Dep.var.: Total leave Coeff. Std.err. Coeff. Std.err. Coeff. Std.err. Coeff. Std.err. Coeff. Std.err.Explanatory variablesPost 1.584 0.612 2.951 1.212 2.768 1.210 2.752 1.202 2.857 1.205Different -5.119 0.834 -3.101 0.910 -2.708 0.911 -1.343 0.927 -1.169 0.948Post*Different 3.541 1.151 3.711 1.145 3.821 1.143 3.405 1.136 3.457 1.140ControlsAge, chldr, wages, local, bigcityYear dummiesOwn level of educationOwn type-level of educationPartner's type-level of educationCouple's type-level edu cellNumber of observations 18077 1807718077

yesyesyes

yesyesyesno

18077

yes

yesno

no

18077yes

nono

no

no no no no

yesnonono no no yes

yesyesnono

(5)(1) (2) (3) (4)




Table 5. Estimation of the effect of economic incentives on mothers’ leave taking

Dep.var.: Total leave Coeff. Std.err. Coeff. Std.err. Coeff. Std.err. Coeff. Std.err. Coeff. Std.err.Explanatory variablesPost 22.220 1.595 25.886 3.055 25.740 3.048 25.929 3.047 20.610 1.547Different 7.265 2.172 9.597 2.096 9.039 2.100 9.107 2.209 6.240 2.254Post*Different -12.118 2.998 -9.035 2.887 -9.029 2.881 -9.335 2.881 -9.392 2.903ControlsAge, chldr, wages, local, bigcityYear dummiesOwn level of educationOwn type-level of educationPartner's type-level of educationCouple's type-level edu cellNumber of observations 1807718077 18077 18077 18077

nono no no no yesno no no yesno no yes yes no

yesno yes no no no

yes yesno yesyesno yes yes yes

(1) (2) (3) (4) (5)




Table 6. Estimation of the effect of economic incentives on fathers’ usage of parental leave

Dep.var.: Parental leave Coeff. Std.err. Coeff. Std.err. Coeff. Std.err. Coeff. Std.err. Coeff. Std.err.Explanatory variablesPost 2.422 1.155 -28.931 2.765 3.008 1.136 -28.512 2.731 -29.118 2.743Different -2.523 0.868 -2.646 1.700 -1.418 0.772 -2.290 1.519 -3.332 1.586Post*Different 3.498 1.092 7.546 2.173Post*CompensationGain (100 DKK) 0.264 0.151 1.327 0.315 2.425 0.576Post*CompensationGainSqr (100 DKK sqr) -0.102 0.045ControlsAge, chldr, wages, local, bigcityYear dummiesOwn level of educationOwn type of educationPartner's type of educationCouple's type-level edu cellNumber of observations 7433

no

yes

nono

18077

yesnono

yes yes

yesyes yes

yesyesyes

yes

(1)

18077 7433

nono

yes yes

7433

yes

no nono

yes

(2) (3) (4) (5)

nono no no no




Results – fathers’ leave usage

The effect of economic incentives inducing fathers to take leave is positive and large

The positive effect tends to be driven by Fathers having their first or second child Fathers employed in local municipalities Long cycle higher education In particular Social Sciences. But also Medium cycle higher education

(Pedagogue/Teacher)




Results – mothers’ leave usage

The effect of economic incentives inducing fathers to take leave is negative and large

The negative effect tends to be driven by Mothers having their third or higher ranked child Mothers employed in state, but also mothers employed in the

local municipalities/counties Long cycle higher education In particular: Social Sciences and Natural and Technical Sciences But also Medium cycle higher education (Pedagogues/Teachers)




Results – validity

Validity of the common trends assumption Education controls -> OK First child -> OK Sector as of year 2001 (pre-reform) -> OK Look at data before 2004 -> not OK Fake reforms

2000 (using pre-reform data only) -> No effect -> OK 2004 (using post-reform data only) -> Positive effect! -> not OK Analyse delayed response, disregard data from year 2002+2003




Dependent variable: Parental leaveCoeff. Std.err. Coeff. Std.err. Coeff. Std.err. Coeff. Std.err. Coeff. Std.err.Explanatory variablesPost 3.976 1.102 -26.164 2.790 -26.057 2.791 -25.958 2.778 -25.731 2.788Different -1.681 0.906 -2.156 1.690 -1.893 1.696 -1.349 1.742 -1.480 1.804Post*Different 5.297 1.310Post*CompensationGain (100 DKK) 3.579 0.692 3.492 0.693 3.327 0.690 3.469 0.694Post*CompensationGainSqr (100 DKK sqr) -0.154 0.051 -0.150 0.051 -0.145 0.051 -0.148 0.051ControlsAge, chldr, wages, local, bigcityYear dummiesOwn level of educationOwn type-level of educationPartner's type-level of educationCouple's type-level edu cellNumber of observations 5736 5736 5736

no nono no

yes yesyesyes

yes

no

(5)

yes yesyes yes

no no13748 13748

yesyesnono

(1) (2)

yes yes yes

nonoyesyes

(3) (4)

no no

yes

Table 7. Estimation of the effect of …… assuming a delayed response




Results – the effect of economic incentives in monetary terms Main specification

Include Post*CompensationGain Post*CompensationGainSqr

Quadratic form, peak at 14.4 days/DKK 1190 Gross income:

For every DKK 82.6 ($11.8) increased wage compensation for transferring a day of leave from the mother to the father

=> fathers supply one extra day of child care Net income:

For every DKK 37.2 ($5.3) increased wage compensation for transferring a day of leave from the mother to the father

=> fathers supply one extra day of child care Elasticity of child care supply = .3

Delayed response Elasticity of child care supply = .43




Approach - career Career outomes

Wage increase DKK, year 1 after birth Wage increase DKK, year 2 after birth Wage increase pct, year 1 after birth Wage increase pct, year 2 after birth Promotion, year 1 after birth Promotion, year 2 after birth Ln wages, year 2006

Career outcome regressed on father’s leave OLS, probit, IV:




Approach - career IV strategies 1) Ideal approach

Post*Different as instrument Validity: OK Strength: OK? (t~3-4)

2) Another approach Post, Different,Post*Different as instruments Strength: OK (due to Post) Validity: ? Identifying assumptions:

Conditional on year dummies, ’Post’ doesnot influence the career outcome Conditional on the individuals’ and the partners’ education, ’Different’ does not influence the career

outcome 3) Alternative approach

Indicators for the couple’s education cell as instruments Strength: OK Validity: ? Identifying assumptions:

Conditional on the individuals’ and the partners’ education, the couples’s exact education cell does not influence career variable




Table 8. Estimation of the effect of father’s parental leave on his partner’s career

Coeff. Std.err. Coeff. Std.err. Coeff. Std.err. Coeff. Std.err. Coeff. Std.err.Outcome variable:

OLS/ProbitFather's parental leave (days) 317.5 20.2 106.8 22.0 0.0013 0.0004 0.0014 0.0004 0.0007 0.0004IV (Post*Different)Father's parental leave (days) 1315.8 1006.9 -837.5 1578.7 -0.0003 0.0198 -0.0063 0.0300 -0.0040 0.0206IV (Post, Different, Post*Different)Father's parental leave (days) 1143.6 522.6 710.9 495.8 -0.0369 0.0110 -0.0293 0.0092 0.0256 0.0117IV (Post*couples type-level edu cell)Father's parental leave (days) 815.5 177.1 372.8 187.3 -0.0052 0.0037 -0.0055 0.0036 0.0119 0.0039ControlsAge, children, local, bigcityYear dummiesOwn type-level of educationNumber of observations

Wage incr., year 1 Wage incr., year 2

yesyesyes yes

Promotion, year 2

yesyes

17933 15105 17985

Promotion, year 1

yes

15151yes

17880

yes

Ln wage in 2006

yes

yes yes yes yesyes




Table 9. Estimation of the effect of father’s parental leave on his own career

Coeff. Std.err. Coeff. Std.err. Coeff. Std.err. Coeff. Std.err. Coeff. Std.err.

Outcome variable:OLS/ProbitFather's parental leave (days) -469.1 21.3 -131.9 25.7 -0.0003 0.0004 0.0004 0.0004 -0.0003 0.0004IV (Post*Different)Father's parental leave (days) 982.0 922.2 417.8 1246.6 0.0137 0.0153 0.0195 0.0205 -0.0557 0.0312IV (Post, Different, Post*Different)Father's parental leave (days) 698.0 878.1 -614.1 1081.1 -0.0003 0.0113 -0.0085 0.0119 -0.0409 0.0187IV (Post*couples type-level edu cell)Father's parental leave (days) -361.5 161.4 208.0 204.1 0.0004 0.0029 0.0012 0.0031 0.0030 0.0030ControlsAge, children, local, bigcityYear dummiesOwn type-level of educationNumber of observations

yes17909 15086 17968 15139

yes yes yes

yesyes yes yesyes yes

Wage incr., year 1

yesyes

Wage incr., year 2 Promotion, year 1 Promotion, year 2

yesyesyes

17848

Ln wage in 2006




Results - careers

Father’s leave usage Improves his partner’s career

Measured by: wage increase in DKK, in pct, log wage in 2006, promotion.

Causal effect for wages, not for promotion Hurts his own career (as expected due to

depreciation) Measured by: wage increase in DKK, in pct, log wage in 2006,

promotion. Unclear whether it is a causal effect Maybe a causal effect for wages, not for promotion




Preliminary conclusion

Economic incentives affect fathers’ leave usage: Increased compensation increases fathers’ leave taking The effect is large:

The couple transfers one more day on average when they gain $5 per day from the father’s parental leave usage (until a max of 14.4 days)

Policy conclusion Unions and employers confederation are on the rigth track in

Denmark by giving individual eligibility to paid parental leave….




Preliminary conclusion

There is an effect of fathers’ child leave on the careers The mother’s career is positively affected by the

father’s leave taking Seems to be a causal relationship at least for wages but not

for promotion

The father’s career is negatively affected Unclear whether it is a causal effect, maybe causal for wages

not promotion

Helena Skyt Nielsen, PhD Professor A A R H U S U N I V E R S I T Y S c h o o l o f E c o n o m i c s a n d M a n a g e m e n t Causes and consequences.

Documents

o f e c o n o

v e r s i t y s c h

r h u s u n

phd professor

day of leave

time leave

parental leave schemes

reform of leave schemes