Identifying Non-Cooperative
Behavior Among Spouses:
Child Outcomes in
Migrant-Sending Households
Session 4E: Growth, Jobs and Earnings
May 15, 2008
Joyce Chen
Ohio State University
• When monitoring is imperfect, allocations can
only be coordinated to the extent that they can
be verified
• Household decision-making may not be fully
cooperative – individuals may attempt to conceal
allocations from each other
• Migration increases the scope for such behavior
in so far as it induces household members to
maintain separate residences
Motivation
• Determine how transparency of income affects
intra-household allocation
– Improve targeting of development programs
– Better understand the consequences of migration on
sending households
• Examine another, less commonly studied
dimension of migration – migration with the
intent to rejoin the sending household
Non-Cooperative Decision-Making
• When transaction costs of cooperation are high,
households revert to non-cooperation (Lundberg
and Pollak, 1993)
• This will not affect provision of household public
goods unless production is organized along
“separate spheres”
• Migration increases cost of monitoring
allocations and forces household production into
a separate sphere
A Simple Example
• A household consists of two decision-makers, a
husband and a wife
• Individuals consume two public goods, x and y
• When the husband migrates, he must rely on his
wife for the provision of public goods
• x is easily observable, but y is difficult to monitor
A Simple Example
• When migration occurs, the husband and wife
can still reach a cooperative agreement on x
• But determination of y is more likely to default to
a non-cooperative process
• Non-cooperative behavior implies
– Shift in consumption towards goods preferred by the
wife, but only if those goods are difficult to monitor
– Magnitude of these changes will be responsive to the
efficacy/intensity of monitoring
Data and Specification
• China Health and Nutrition Survey
–
–
–
–
5 rounds (1989, 1991, 1993, 1997, 2000)
approx. 4000 households and 15,000 individuals
Observable outcomes – height, weight, schooling
Less observable inputs – individual time allocation,
nutritional intake
• 1990s were a period of rapid growth in intranational migration in China, fueled by relaxation
of migration restrictions and increased openness
and marketization
Data and Specification
• Migration defined as living away from home for
at least one month in the last year
• Most outcomes defined over the previous one
week – restrict sample to dads away during the
entire week preceding the survey
• To distinguish non-cooperative behavior from
income effects, control for full income with
wages and productive assets
Data and Specification
• Migration is endogenous - individual fixed effects
included to control for unobserved
characteristics correlated with migration
• Community-year fixed effects included to control
for time-varying factors correlated with migration
• Estimate reduced-form demand equations for
schooling, health and household labor
Data and Specification
• Include number of months away
– Proxy for intensity of monitoring
– Outcomes may require time to adjust
• Allow child’s age and number of sibling, by
gender, to vary with father’s migration status
• Additional controls for parents’ ages, household
size and month and year of survey
Identifying the Counterfactual
• Direct effects of migration
– Reduction in father’s household labor
– Increase in household income
– Possible change in bargaining power
• Appropriate counterfactual is the set of
allocations that would have been chosen,
conditional on these changes, if spouses could
costlessly commit to cooperation
Identifying the Counterfactual
• Reduction in father’s household labor: increases
mother’s household labor, ambiguous effect on
child household labor
• Increase in household income: increases
mother’s household labor and decreases child
household labor
• Increase in father’s bargaining power: also
increases mother’s household labor and
decreases child household labor
Distinguishing Non-Cooperation
• Migration in a cooperative model should
increase mother’s household labor, unless
migration increases mother’s bargaining power
• Non-cooperative model suggests that the
mother will decrease own household labor when
the father migrates because time allocation is
difficult to monitor
Table 2. Mothers' Time Allocation,
Mother Fixed Effects Estimates
Do Any Work Hours
Chores (ex. chores)
Father Away
0.077 *
12.19
(1.66)
(1.52)
Months Father Away
-0.031
-5.581 *
(1.63)
(1.81)
Months Away Squared
0.002
0.459 *
(1.48)
(1.90)
Marginal Effect of Away
-0.041
-4.397
(1.36)
(1.03)
Sample Mean
Observations
0.974
6450
43.50
5996
• Mothers’ labor, both in the household and in income-generating
activities, is decreasing in months the father is away
• Decrease in mother’s household labor is not consistent with a
model in which decision-making is cooperative and time allocation
adjusts to compensate for the father’s absence
Table 3. Children's Time Allocation,
Child Fixed Effects Estimates
Prepare
Do
Food
Laundry
Boys
Girls
Boys
Girls
Dad Away
0.253
-0.116
0.228
-0.365
(1.51)
(0.56)
(1.17)
(1.27)
Months Away
-0.047
0.065
-0.084 *
0.164 **
(0.97)
(1.01)
(1.65)
(2.11)
(Months Away)^2
0.003
-0.003
0.007 *
-0.013 **
(0.82)
(0.63)
(1.79)
(2.19)
(Age-6)*Away
-0.044
0.006
0.013
-0.015
(1.15)
(0.11)
(0.31)
(0.21)
(Age-6)^2*Away
0.004
0.000
-0.001
0.002
(0.95)
(0.07)
(0.35)
(0.25)
Marginal Effect of Away
-0.064
0.120
-0.061
0.191
(0.74)
(1.07)
(0.07)
(1.56)
Sample Mean
Observations
0.057
0.117
8476
0.073
0.187
8329
• Probability that girls do either chore is increasing in months away,
and the opposite for boys
• Statistically significant only for laundry, but average marginal effects
are quite large
• The level of household production is observable,
but the inputs to production are not – substitute
children’s time for mom’s
• Results may be consistent with a cooperative
model of the household if migration increases
mothers’ bargaining power
• Should lead to changes in other goods favored
by mothers, e.g. child health (Duflo, 2003;
Thomas 1990)
Table 4. Children's Health and Nutrition,
Child Fixed Effects Estimates
Body Mass
Daily Calorie
Index
Intake
Boys
Girls
Boys
Girls
Dad Away
-0.671
0.657
288.0
-409.8
(0.80)
(0.38)
(1.04)
(0.92)
Months Away
0.203
-0.18
10.96
53.31
(0.80)
(0.40)
(0.12)
(0.38)
(Months Away)^2
-0.025
0.02
-3.574
-2.433
(1.13)
(0.52)
(0.43)
(0.22)
(Age-6)*Away
0.155
-0.539
-181.0 **
211.4 *
(0.60)
(1.03)
(2.05)
(1.76)
(Age-6)^2*Away
-0.015
0.063
18.11 **
-24.68 **
(0.61)
(1.21)
(1.98)
(1.99)
Marginal Effect of Away
0.351
-0.898
-139.7
160.5
(0.96)
(1.05)
(0.98)
(0.70)
Sample Mean
Observations
17.14
17.11
6121
1851
1711
7303
• No significant effects on child health - not consistent with an
increase in mother’s bargaining power
• But large changes in calorie intake - changes in time allocation must
be balanced with changes in nutrition in order to maintain health
• Changes in time allocation are larger for
– Children who are more productive
– Activities that provide higher disutility to mothers
• Siblings have a reinforcing effect on the
extensive margin and an offsetting effect on the
intensive margin
• Mothers’ BMI appears stable
– In fact, mothers consume less calories and protein to
conceal the reduction in their household labor
– Suggests that non-cooperative behavior occurs in
equilibrium because an incentive-compatible contract
must give moms higher consumption of private goods
Conclusion
• Mothers do engage in non-cooperative behavior,
in a surprising way
– Not consistent with change in bargaining power
– Not consistent with reallocation of time to compensate
for father’s absence
• But, in this case, it is largely innocuous - does
not appear to affect children’s human capital
accumulation
Further Research
• Need data on both migrants and sending
households to provide a complete picture of
migration and its economic impact
– Importance of transparency, monitoring and
enforcement… for both parties
– Policies that affect migration decisions and
circumstances
• Improve data collection for temporary migrants,
e.g. income vs. remittances, travel and visitation
patterns, individual expenditures