1
Investments in Children’s Human Capital
Robert A. PollakWashington University in St. Louis
December 4, 2009
2
Big Differences
Big differences in the amounts that different families invest in their children.
Big differences in children’s education outcomes:
test scores years of education; HS graduationWhat does economics have to say about these
differences?
3
Implicit Assumptions in Mainstream Economics - 1
Focus on family investments.Parents as traditional nuclear family: married when the children are born and stay
married to each other.Becker Treatise on the Family, 1981, 1991Behrman, Pollak, and Taubman, From Parent to
Child, 1981, 1991.
4
Family Structure
DivorceNonmarital fertility
70% of non-Hispanic black births nonmarital 48% of Hispanic births nonmarital
25% of white births are nonmarital
5
Outcomes for Children
Children do better if they come from traditional nuclear families
Family structure correlated with mothers’ educationChildren do better if their mothers have more
educationIt is not clear that any of this is causal: Would children
do better if their unmarried mothers married their unmarried fathers?
6
Two References
• McLanahan, Sara, "Diverging Destinies: How Children Fare Under the Second Demographic Transition," Demography, Vol. 41, No. 4, (November 2004), 607-627.
• Lundberg, Shelly and Robert A. Pollak, "American Family and Family Economics," Journal of Economic Perspectives, Vol. 21, No. 2, (Spring 2007), 3-26.
7
Implicit Assumptions in Mainstream Economics - 2
Parents make the decisions: parents active; children passive decisions about preschool
(Some models in which children are active; decisions about medical school).
Alternative: “bargaining” between parents and children
Agency for children
8
Explanations for Differences in Investments in Children - 1
Why do we see such big differences in the amount invested in children from different families?
Why do low SES families invest so little?Why do high SES families invest so much?(Fertility as background question: Why have
children at all? Below replacement fertility in Europe and Japan.)
9
Explanations for Differences in Investments in Children - 2
* Differences in “Preferences”* Differences in Constraints
Productivity of time (if you can’t do math, you can’t help child with math)
Availability of time/ value of time/ wage Money (to buy market substitutes)
10
Other Family Members: For example, Grandparents
Are they around? Compton and Pollak address this. The answer, for many families, is yes.
Do they play a role? We don’t know. The literature in economics and sociology pays little
attention to grandparents except when grandparents live with the grandchild and/or are primary caregivers. See Dunifon (JM&F, 2007)
Proximity: suppose grandparents live close to (e.g., within 30 miles) but not with grandchildren
11
Adults Living Near Parents
Janice Compton and Robert A. Pollak, "Proximity and Coresidence of Adult Children to their Parents: Description and Correlates," University of Mich Retirement Research Center, WP 2009-215, October 2009.
Americans live surprisingly close to their parents.We focus on mothers, but for mothers and fathers who
live together, it doesn’t matter.We consider mothers who are Alive and Living in the
United States (ALUS)
12
MEDIAN DISTANCES
Including coresidents (d≥0)
Excluding coresidents(d>0)
Unmarried Adults
Men 5 miles 15 miles
Women 8 miles 15 miles
Married Adults
Her Mother 20 miles 22 miles
His Mother 25 miles 25 miles12
Adults (≥25) with Mothers ALUS
13
Why Do Adult Children Live Close to their Mothers? - 1
Who benefits from proximity?What is the relationship between proximity and transfers of money? Of services (“time transfers”)?Adult children may benefit from close proximity and time transfers when they are young and have young children.Mothers may benefit from close proximity and time transfers when they are elderly, disabled, and widowed.
13
14
Why Do Adult Children Live Close to their Mothers? - 2
We assume that the financial and psychic costs moving are high enough that proximity should be analyzed within a life-cycle framework.
A life-cycle framework implies that we cannot understand proximity by looking at the current needs.
We discuss this “why” question and the “who benefits” question later.
.
14
15
Most Americans Marry Someone Born in the Same State
15
Percentage of Couples Born in the Same State
All Couples
2000 59.21990 58.91980 61.1
U.S. Census (IPUMS) data. Sample includes all couples in which both spouses are 25 years of age or over and born in the U.S.
1616
Education and Couples’ Birth State
Table 2: Couples’ Birth State
Low-Power Couples
Part-Power: She has college degree
Part-Power: He has college degree
Power Couples
All
Percentage of Couples from same birth state
2000 64.1 56.8 54.0 45.9 59.2
1990 62.8 54.8 52.5 45.1 58.9
1980 64.2 55.8 53.1 47.9 61.1
Percentage of Sample
2000 62.4 8.5 12.3 16.8 100.0
1990 69.3 6.2 12.0 12.5 100.0
1980 75.6 4.1 11.2 9.1 100.0
U.S. Census (IPUMS) data. Sample includes all couples in which both spouses are 25 years of age or over and born in the U.S.
17
Proportion of Couples Born in the Same State
0
10
20
30
40
50
60
70
1980 1990 2000
Low Power Couples Part-Power: She has College DegreePart-Power: He has College Degree Power Couples
18
Proportion of Couples Born in the Same State
0
10
20
30
40
50
60
70
1980 1990 2000
Low Power Couples Part-Power: She has College DegreePart-Power: He has College Degree Power Couples
9.1 12.5 16.8
19
Data: NSFHNational Survey of Families and HouseholdsWave 2 (1992-1994)We use the adult child and his or her spouse as unit
of observation– Distance to his mother, distance to her mother– Transfers to and from both mothers: time transfers
and monetary transfersMost previous studies used the mother and her
children as unit of observation.Migration: of adult child; of his mother; of her
mother. We plan to look at this.
19
20
Proximity as Active Research Topic
Conference at University of Michigan: “Proximity, Coresidence and Intergenerational Transfers” May 2009
20
2121
Proximity and Coresidence - 1If distance were the right metric, coresidence would be the limiting
case of close proximity. But distance isn't the right metric. Living next door vs coresidenceTheoretical:
discontinuity in costsdiscontinuity in privacy
Empirical: Data do not support treating coresidence as limiting case of close
proximity.
22
Proximity and Coresidence - 2
Implications for empirical work:Use Multinomial logit/probitDo not use Tobit, ordered logit/probit
–They assume one set of regressors–They may give misleading results
22
23
Correlates of Proximity and Coresidence
Coresidence: Very few married couples live with either his mother or her mother.
Proximity: Strong education gradient: couples with more education live farther from his mother and her mother.
Age gradient: older children live farther awayTime transfers: close proximity associated with
more time transfers (in both directions).
23
2424
Education, Proximity and Transfers
Education is a strong predictor of proximity:Adults with a college degree are much less
likely to live near their mothers.
2525
Some Facts about Education and Proximity
• Strong Education Gradient
Low-Power Couples
Part-Power: She has college degree
Part-Power: He has college degree
Power Couples
NSFH Wave 2. Lives within 30 miles ofNeither mother 18.9 25.5 35.8 49.4Her Mother 17.4 23.2 13.6 15.9His Mother 13.9 16.4 15.3 16.3Both Mothers 49.9 34.8 35.3 18.4
2626
More Facts about Education and Proximity
• Strong Education Gradient
Low-Power Couples
Part-Power: She has college degree
Part-Power: He has college degree
Power Couples
2000 U.S. Census. Lives in birth state ofNeither 22.6 26.1 36.2 40.1Hers 12.1 12.8 13.3 13.5His 11.9 14.4 11.8 13.9Both 53.3 46.7 38.7 32.5
2727
Education, Proximity and Transfers - 1
Education is a strong predictor of proximity– Adults with a college degree are much less
likely to live near their mothers.– College educated more likely to migrate
(wider job market, may have gone to college away from home).
– Less likely to need transfers.
2828
Education, Proximity and Transfers - 2
Education is positively related to transfers of both time and money. – Adults with a college degree are much more likely to give
and receive transfers from their mothers.– Selection – since college educated more apt to move away
from home, those who stay are those who are more likely to give/receive transfers.
– Strategic behavior of mothers – knowing that college educated have greater opportunities elsewhere, may transfer more to them to ‘bribe’ them to remain.
2929
Next Project
Labor supply and Intergenerational Transfers– Focus on women– Labor supply and proximity
• Can use proximity rather than transfers• Impact of close grandchildren on retirement• Impact of close grandmothers on labor supply
of daughters.
3030
Education, Proximity and Transfers - 3
Education of mothers has an independent effect on proximity. – Adult women are less likely to live near mothers with a
college degree. – Mothers more likely to have migrated away from home
herself - less extended family around– Demonstration effect– Mothers better able to identify and obtain market
resources for her care. May have less need for non-market family care.
– Mothers more likely to be in the labor force, may be less likely to provide child care.
31
ARE FAMILIES PARETO EFFICIENT?
31
32
What Can We Learn from Family Migration?
33
Family vs Household vs Individual
• The economics of the family beyond the household
• "household production" • “Treatise on the Family”• Ellickson, The Household: Informal Order
around the Hearth, 2008
3333
34
What We Want to Explain
• Migration and Location• Proximity and Coresidence• Who lives with whom is endogenous• Who lives near whom is endogenous
34
35
Long-term Care of Disabled Elderly Parents
• Which child will be the primary caregiver? Engers and Stern (2002 IER)cooperative game and two-stage game
• Coresidence with disabled elderly parentPezzin, Pollak, and Schone (2007 CESifo
Studies) Three-person, two-stage game: children decide whether to invite coresidence; parent decides which invitation to accept
3535
36
Modeling Family Migration
• Players • Feasible Set• Preferences• The Game
3636
37
Players
• Two-person vs three-person vs n-person games• Coalitions• Boundaries of the family not sharply defined:
– “extended family” • Family trees are bushy • Bernheim and Bagwell (1988 JPE)
"Is Everything Neutral?"
3737
38
Feasible Set
• Coresidence• Proximity
– Residential location vs – Cities as locations
3838
39
Preferences
• Why care about location?– Wages, prices, amenities, friends, family
• Why care about proximity?long-term care; child care;proximity to grandchildren;my sons’ locations affects my utility
• "Externalities" within the familyinterdependent preferences
• Proximity as a "family public good"
3939
40
Specifying the Game
• Describing vs inventing the game• Cooperative vs noncooperative games• Konrad et al. specify noncooperative game• Without binding agreements, two-stage games
likely to yield Pareto inefficient equilibria
side payments?binding agreements?dynamics
4040
41
Norms
• About caregiving responsibilities• About coresidence and proximity
4141
42
Norms
• Distribution• Pareto efficiency• Property rights interpretation: Coase• If you are the child whom norms designate as
responsible for providing care, is it ok for you to pay your sibling to provide care?
• Law vs norms: Coase vs Ellickson• Ellickson (1991 Order without Law: How
Neighbors Settle Disputes)
4242
43
Remaining Work
Sorting out causal relationship from observational data.Which adult children are in close to parents is
endogenous.What is the effect on outcomes for children (e.g.,
education) of being in close proximity to grandparents.
Controlling appropriately for parents’ educational attainment, is being in close proximity to grandparents protective for children?