Seasonal Migration and Early Childhood Development in Nicaragua Karen Macours (SAIS-Johns Hopkins) and Renos Vakis (World Bank)

Seasonal Migration and Early Childhood Development in Nicaragua

Karen Macours (SAIS-Johns Hopkins) and

Renos Vakis (World Bank)

2

Motivation Evidence of remittances and permanent

migration on: Income: McKenzie et al. (2006) + Taylor et al.

(2005) + Poverty: Lopez-Cordova and Olmedo (2006) + Schooling: Yang (2006): +, McKenzie and Rapoport

(2005): - Health expenditures: Amuedo-Dorantes and Pozo

(2005) + Infant mortality: Hildebrand and McKenzie (2005) – Birthweights: Duryea et al. (2005) +

3

Motivation Still, most studies either explicitly focus

on permanent migration or do not distinguish seasonal from permanent migration, even if both determinants and effects might be very different

Seasonal migration Traditionally important strategy for both

income diversification and risk coping in many developing countries

South-south movements becoming very important

4

Motivation Seasonal migration and human capital accumulation

Temporary absence of parents can have potential important consequences for the cognitive development and long-term human capital accumulation of children left behind

Emerging literature on early childhood development (ECD) emphasizes the role of parenting (e.g. Behrman, Cheng and Todd, 2004; Gertler and Fernald, 2004; Paxson and Schady, 2006)

Given that absence of parents is arguably the most extreme form of lack of parenting, this would suggest a potential strong negative effect of migration on early childhood development

Little is understood about potential trade-off of seasonal migration between parenting (stimulation) and income (nutrition, health)

5

Motivation

Determinants/Correlates of Cognitive Development Nutrition Stimulation Home inputs Micro-nutrients Health Parent’s socio-economic status Pre-school programs

=> Impact of seasonal migration a priori unknown

6

Motivation Nicaragua context

Border region in North of Nicaragua Shock-prone area with mainly subsistence agriculture

(corn and beans) 50 % of households rely on seasonal migration to

complement and diversify income: 20% of household income Fathers (41%), mothers (8%) and other members (13%)

migrate Children typically stay behind (in care of family) On average 3 months Majority to other countries in Central America (56%)

Focus on children of pre-school ages for whom direct parental care and stimulation is arguably the most important

Permanent migration (appears to be) much more limited

7

Research questions (and answers)

1) Are children of seasonal migrants disadvantaged in terms of ECD outcomes?

Yes

2) Is there a trade-off between increased migration income and temporary absence of household members for children’s ECD outcomes?

Mothers’ seasonal migration has a positive effect on ECD outcomes No significant effect for other household members (including fathers)

3) What are the mechanisms and channels by which mothers’ seasonal migration protects children’ ECD outcomes?

Income gains: change in intra-household resource allocation and bargaining position of mothers inducing higher investments on children (e.g. nutrition and health)

Stimulation potentially compensated by alternative care (abuelita, ECD programs)

8

Methodology1) Correlations between migration and ECD

• Cognitive development : TVIP• Anthropometric measures

2) IV estimates to analyze relationship between migration and ECD by exploring seasonal migrant patterns

3) Explore alternative hypotheses to understand the trade-off between income and stimulus

9

Data More than 4000 households from 6

municipalities, targeted for poverty and vulnerability to droughts (part of a CCT pilot)

Information on 1800 children (out of which 1500 are between 3-7 years) whose parents are the household head

Outcome indicators TVIP (Spanish version of PPVT) for children 3-7

Tests of receptive vocabulary (recognition of pictures of objects and situations based on common and gradually less common stimulus words

Predictor of income and wages Standardized on Spanish-speakers in Mexico and Puerto

Rico Alternatively: internally standardized

Anthropometrics (for children up to 5)

10

Q1. Are children of seasonal migrants disadvantaged in terms of ECD outcomes?

6080

100

120

140

TVIP

sta

ndar

dize

d sc

ore

40 50 60 70 80Age in months

No seasonal migrants Seasonal migrants

bandwidth = .8

.

Figure 1 – TVIP scores by age and seasonal migration status

11

0.2

.4.6

.81

% s

tun

ted

0 20 40 60age in months

No seasonal migrants Seasonal migrants

bandwidth = .8

.

Q1. Are children of seasonal migrants disadvantaged in terms of ECD outcomes?

Figure 2 - % of children that are stunted: by seasonal migration status

12

Q2. Is there a trade-off between increased migration income and temporary absence of household members for children’ ECD outcomes?

Explore heterogeneity in seasonal migration patterns

Type of seasonal migrant

Mother Father Other member Any household member

Households with migrants (%) 8* 41 13 50 % of seasonal migrants who go abroad 45* 59 39 56 Migration duration (days) 85 86 1401 1221 Migration income Total earned (Cordobas) 6976 6028 5435 5652 Brought back (Cordobas) 4174* 3356 2694 3115 % of migration income brought back 73* 65 57 64 % brought back as share of total hh income 23 21 15 19

13

Q2. Is there a trade-off between increased migration income and temporary absence of household members for children’ ECD outcomes?

Instrument seasonal migration with household shocks 48% of mother, 23 % of fathers, and 37% of

others that migrated do not migrate every year Shocks are correlated with seasonal migration What about the exclusion restrictions?

Over-identification test Additional controls Would bias the results downwards (e.g. shocks might

negatively effect ECD through direct effect on consumption)=> Underestimation of seasonal migration effect

14

Q2. Is there a trade-off between increased migration income and temporary absence of household members for children’ ECD outcomes?

TVIP standardized score (1) (2) (3) (4) (5) Total # of days - Mother migrated 0.276** 0.238* 0.282** 0.252* 0.884 (1.97) (1.69) (2.05) (1.71) (0.53) Total # of days - Father migrated -0.015 -0.006 0.008 0.010 0.052 (0.22) (0.09) (0.06) (0.14) (0.22) Total # of days - Other household member migrated -0.071 -0.073 -0.116 -0.063 -0.200 (1.07) (1.16) (1.54) (1.08) (0.44) Controls for individual, household and community characteristics yes yes Yes yes yes Just identified no yes Yes yes yes Excludes household with permanent migrants no no Yes no no Excludes households with other female adults no no No yes no Includes any child in household no no No no yes Observations 1534 1534 1367 1221 2080

Joint F-test of exclusion restriction – Mothers 4.13 4.37 7.15 6.37 1.56

Joint F-test of exclusion restriction - Fathers 4.01 5.33 3.44 3.74 5.72

Joint F-test of exclusion restriction - Other 5.69 6.86 9.05 12.24 2.39

Anderson instruments validity (chi2) 5.997 5.997 2.611 5.719 0.262

P-value 0.0499 0.0143 0.1061 0.0168 0.6085

Sargan over-identification (chi2) 0.612 -

P-value 0.4342 -

Instruments for column (1) include households in last 12 months: plagues, wages, adult illness and price shock. Column (2) to (5) exclude price shocks.

Absolute value of z statistics in parentheses

* significant at 10%; ** significant at 5%; *** significant at 1%

Table 4: Seasonal Migration and ECD: IV estimates

15

Q3. What are the mechanisms and channels by which mothers’ seasonal migration protects children’ ECD outcomes? Income, empowerment and intra-household

bargaining Seasonal migrant women contribute more to household

income Intra-household bargaining might be shifted towards

women Improve investments on children’s human capital (e.g.

through better nutrition and health) Supporting evidence 1

Non significant results when we include all children in our sample, irrespective of whether they are children of the household head (Table 4, specification 5)

Suggests that seasonal migration decisions in nuclear households (and to that extent intra-household resource allocation) driven by different factors than in extended (often multigenerational) households

16

Q3. What are the mechanisms and channels by which mothers’ seasonal migration protects children’ ECD outcomes? Supporting evidence 2

TVIP scores are positively correlated with household consumption The relationship is stronger for households with seasonal migrant mothers Consistent with increased women’s empowerment and higher investment in children

60

80

100

120

140

TV

IP s

tan

dard

ize

d sc

ore

0 5000 10000 15000 20000pc consumption aggregate

Non-migrant mothers Migrant mothers

bandwidth = .8

.

Figure 4 – TVIP, Seasonal Migration, Mothers and Wealth

17

Q3. What are the mechanisms and channels by which mothers’ seasonal migration protects children’ ECD outcomes? Stimulation and alternative childcare

Existing childcare within the household? important to account for family members that do not

migrate, and might become responsible for care-giving during the migration episodes

effect of migration itself might depend on whether childcare is taken over by other adult household members who might provide similar levels of parenting and/or stimulus

Specification 4 (Table 4) excludes all children from households where there is an adult female household member

No change in results, suggesting that having an alternative caregiver in the household might not be key for ECD reinforcing effect of mothers

18

Q3. What are the mechanisms and channels by which mothers’ seasonal migration protects children’ ECD outcomes? Stimulation and alternative childcare

Childcare outside the household - the “abuelita” story from field observations Mothers typically leave the children in care of a grandmother or

other family member The children become temporary members of the other household Since the cash income of the other household is not affected by

the migration, access to nutrition and care might be more constant

Fathers’ seasonal migration Temporary income shock Mother’s care might decrease if she is forced to look for a cash

income during the father’s absence Social norms prevent mothers to place children in another

household during the father’s absence when mothers are still present in the community.

Our data do not allow analyzing these hypotheses

19

Q3. What are the mechanisms and channels by which mothers’ seasonal migration protects children’ ECD outcomes? Income and stimulus as complementary

inputs Stimulus might be important for early

childhood development, but only if the child has reached a minimum nutritional threshold

Given high levels of malnutrition, lack of stimulus might not have strong effects as long as nutrition is severely constrained (Figure 4)

Seasonal migration income of mothers might then be a pre-condition for early childhood development through stimulus

20

Q3. What are the mechanisms and channels by which mothers’ seasonal migration protects children’ ECD outcomes? How to test this hypothesis formally?

Data includes information on an ECD and stimulation program (PAININ) whose specific goal is to provide stimulus to young children

39 % of the pre-school children in our sample have participated in PAININ

Information on program placement (targeting rules) and determinants of self-selection

Apply matching methods to measure impact of program and separate income from stimulation effects

21

Q3. What are the mechanisms and channels by which mothers’ seasonal migration protects children’ ECD outcomes?

6080

100

120

40 50 60 70 80 40 50 60 70 80

2 lowest consumption quintiles 2 highest consumption quintiles

Non-Painin Painin

TV

IP s

tand

ard

ized

sco

re

Age in months

bandwidth = .8

Striking complementarity between this stimulus program and household consumption TVIP scores of children from households in PAININ higher than those not Differences much smaller for children in the lowest consumption quintiles (nutrition constraint binding)

22

Q3. What are the mechanisms and channels by which mothers’ seasonal migration protects children’ ECD outcomes?

60

80

100

120

40 50 60 70 80 40 50 60 70 80

Non-migrant mothers Migrant mothers

Non-Painin Painin

TV

IP s

tan

dard

ize

d sc

ore

Age in months

bandwidth = .8

Stronger if we look at the relationship between Painin and TVIP scores for children of seasonal migrant and non-migrant mothers

23

Summary Negative correlation between ECD and migration After instrumenting seasonal migration with exogenous shocks

Mother’s seasonal migration has a significant positive effect on ECD outcomes, suggesting that the positive income effect might be larger than potential lack of parenting effect

No significant effects of fathers’ or others’ migration on ECD Exploring the income versus stimulus trade-off

Income effect driven by changes in intra-household resource allocation and bargaining

Stimulus effect may be partially underestimated by “abuelita” effect

Potential complementarities between income and stimulus Stimulus less important at low levels of malnutrition Seasonal migration income of mothers might be a pre-condition for

ECD through stimulus

24

Implications

On migration Intra-household dynamics imply that social costs

related to seasonal migration by women are not ex-ante obvious

Seasonal migration by women not necessarily bad Food for thought for policy makers

On early childhood development Women’s control over cash income might be

important Potential importance of targeting cash transfers

to women

25

Next steps

Explore further intra-household bargaining ongoing randomized evaluation will allow

addressing some of these ideas magnitude of cash transfer similar to cash

brought back initial IE results on empowerment consistent

(e.g. intra-household decision making etc) …

26

Instruments? First stageTable 5: First stage regression for IV estimates (Table 4, specification 1)

Total # of days mother migrated

temporary

Total # of days father migrated

temporary

Total # of days other household member migrated

temporary

Wage shock -3.195 26.765*** -1.002

(0.59) (2.77) (0.08)

Adult illness shock 12.647*** 19.896*** 41.093***

(3.10) (2.76) (4.53)

Price shock 7.560* 1.693 13.488

(1.84) (0.23) (1.48)

Agricultural plague shock -3.305* 0.289 2.651

(1.88) (0.09) (0.68) Controls for individual, household and village characteristics Yes Yes Yes

Observations 1534 1534 1534

R-squared 0.10 0.09 0.12

Absolute value of t statistics in parentheses

* significant at 10%; ** significant at 5%; *** significant at 1%