Upload
ralph-stewart
View
219
Download
1
Tags:
Embed Size (px)
Citation preview
Childcare availability and female labor supply
Anna Lovasz - Agnes Szabo-Morvai
The impact of day-care services on mothers’ employment, fertility, and redistribution in
Visegrad countries - WorkshopBudapest, March 30-31, 2012
Research question and literature
• How does the lack of formal childcare availability constrain female labor supply?– International evidence that it does constrain:
Apps&Rees 2001; Kimmel 1992,2001; Lokshin 2004– Who is most affected by constraint?
• By income, education level, region/settlement type, family status, age
– Is the market for private daycare „stepping in” where public is insufficient?
• Is this increasing inequality based on affordability?
Relevance• Policy issues:
– Where to build kindergartens?– Who should pay and how much for nurseries?– Should market for private daycare be encouraged more
(decrease administrative barriers, etc)?Labor market activity
– Bick, 2010: lack of subsidized childcare is a barrier to female labor supply
– Connelly, 1992: higher child care costs are the primary reason of lower participation rate of mothers
Fertility– Apps & Rees, 2001, Del Boca and Sauer, 2009 : countries with
better prospects for mothers of small children (availability of childcare and flexible jobs), have higher female labor supply and fertility rate
Childcare availability
DataCombine three data sources, Hungary 2002-2011:• Labor Force Survey
– Household composition, labor status, children– Rotating panel, at most 6 quarters’ data about one
household• T-STAR Geographical data
– Nursery and kindergarten availability, family daycare (2008-2010), commuting
– Matched to LFS using settlement codes• Wage and Employment Survey
– Expected wage according to education, industry, etc.
Childcare scarcity in Hungary
Scar
city Kindergarten: 69%
Nursery: 99%
Utilization rate = enrolled children / available places
Methodology: what happens at age 3?
• Increase in availability between nursery and kindergarten effect on LS?– Kindergarten should accept all children above 3 if
open places left– Largest enrollment wave in September – Continuous enrollment if unfilled places
• typically in lower quality kindergartens• often wait until next September, when kids leave for
school
• Problem: other effects at age 3– Maternity leave ends– Willingness to separate from child?
Factors affecting childcare usage and mother’s labor market participation when child turns 3
Childcare availability
Willingness to separate (Blaskó)-This factor is present and has a strong effect-Its timing is uncertain-Continuous variable
Maternity leave-High-sum maternity support ends at age 2, no work allowed-Low-sum maternity support (~ 100 EUR) ends at age 3-Mothers are allowed to work and receive low-sum support
Facts and Figures I.0
.2.4
.6.8
Act
ivity
ra
te
0 2 4 6 8Age of youngest child (year)
Activity rate vs age of youngest child
Facts and Figures II..5
.52
.54
.56
.58
Act
ivity
ra
te
0 5 10 15Month of the year
Activity rate through the year (child: 3-3.5y)
Facts and Figures III.0
10
00
20
00
30
00
40
00
Co
un
t
0 2 4 6 8Age of youngest child (year)
Want a job: NO; N/l b/c child: NO Want a job: NO; N/l b/c child: YESWant a job: YES; N/l b/c child: NO Want a job: YES; N/l b/c child: YESWork
Working and reasons for not searching
Work
Don’t want; N.l. b/c childcare problem
Don’t want; N.l. b/c NO childcare problem
Facts and Figures IV.0
.2.4
.6.8
Perc
ent
0 2 4 6 8Age of youngest child (year)
Work Available for work in 2 weeksNot looking for a job b/c of a child Not search, but want a job
Work availability and search
Working
Available
Not looking b/c of child
Not looking, but want
Ideal experiment and problems• Population of women who want a child (unobservable)• Assign children to them randomly (no sample selection)• Randomly offer them (group 1) or not (group 2) childcare
(childcare availability is exogenous) Compare the activity rate of group 1 & 2• Problems in real life data:
– Selection into motherhood– Endogeneity of childcare availability– Concurrent „treatment”: end of maternity leave
• Usually tackled by parametric, multi-equation models– Selection into motherhood is usually not handled by these
We plan to take an approach that requires less behavioral assumptions but handles these problems
Quasi-experiment: regression discontinuity design
• Random assignment would solve selection problem
• Can think of mothers of children aged 2.7-3.3 as very similar, except:– Under 3: only nursery, low childcare availability (7%
on average)– Over 3: kindergarten, high availability (83% on
average)• In this „discontinuity sample”, assignment is
random– Child age not correlated to characteristics that
determine participation– Except: willingness to outsource daycare
Strategy 1
Kr: regional kindergarten availability: available kg places / number of children (or # of chilod-bearing age women)Nr: regional nursery school availabilityGamma i: other parameters that affect availability
Local Average Treatment Effect
Age of youngest child
Activ
ity ra
te
3
LATE
Observed
Unobserved
Preliminary results: activity rate by level of change in childcare availability
• Availability: number of places / number of children in population of given age • Change in availability if: No nursery, but kindergarten available OR availability of
kindergarten is higher0
.2.4
.6.8
Act
ivity
ra
te
0 2 4 6 8Age of youngest child (year)
(mean) aktiv_change (mean) aktiv_nochange
Activity rate vs age of youngest child
Strategy 2
• Exploit gap between when child turns 3 (end of maternity leave) and kindergarten enrollment month (mostly in September)?
Maternity leaveEnrollm. 0 1 Total
0 3,134 55,468 58,602 1 59,266 0 59,266
Total 62,400 55,468 117,868
Preliminary results: activity by regular or late enrollment
0
.2.4
.6.8
Act
ivity
ra
te
0 2 4 6 8Age of youngest child (year)
(mean) aktiv_enrolltime (mean) aktiv_enrolllate2
Activity rate vs age of youngest child
Strategy 3
• Available places in 2010: – in nurseries : 26.000– in family daycare: 4.000
• appr. 15% increase in available places since 2007, with geographical differences
• Source of variation: – geographical and time differences of childcare
availability– regional differences in availability growth
Issues/questions• Develop model and RD design: what is treatment? Exogenous change in change in availability (Ex: retirement of
kindergarten teacher leads to closing) Reduced form: we observe childcare availability and labor
market participation, but do not observe actual enrollment for given mothers
• Female labor supply or household decision model? literature shows decisions made jointly when young children
present (Lundberg 1988)• Fertility decision not modeled• Include family members: informal childcare• Childcare availability or affordability?• Availability at location: living or working? use Kertesi et al.: composed small regions based on
commuting data• Availability of flexible jobs?
ANY COMMENTS ARE WELCOME!Thank you for your attention,