Childhood Migration and Educational Attainment - Tables

Figures N SEC Figure 1: Histogram ∆JU , Sample of 15 - 20 Year Olds podc N SEC for all migrants age 15 - 20. ∆JU N SEC = ȳ Note: Figure shows the distribution of location quality change ∆JU pdc − ȳpoc , where ȳpdc podc podc and ȳpoc are average junior secondary school graduation rates of permanent residents in destination district d and origin district o, respectively, of birth cohort c and parent education group p. Data source: 2000 and 2010 Indonesian Population Census. Figure 2: Location Exposure Effect on Junior Secondary School Graduation (Age 15 - 20) (a) β̂m (b) Linear Fit of β̂m N SEC on junior secondary school graduation by age at migration during Note: Panel (a) shows the effect of location quality change ∆JU podc childhood (with 95% confidence intervals, standard errors clustered two-way on origin and destination district), Panel (b) shows the linear fit of those regression coefficients, the location exposure effect. The regression in Panel (a) controls for origin × birth-cohorts fixed effects, age-atmigration × parent-education group fixed effects, and highest completed degree of the household head interacted with gender of household head, and highest completed degree of their spouse interacted with the spouse’s gender (Equation 4). Maximum age at migration is 14 years. Sample: one-time migrants age 15 - 20 from 2000 and 2010 Indonesian Population Census. Figure 3: Location Exposure Effect on Years of Education (Adults Age 21+) (a) β̂m (b) Linear Fit of β̂m R Note: Panel (a) shows the effect of location quality change ∆Y odc on years of education by age at migration during childhood (with 95% confidence intervals, standard errors clustered two-way on origin and destination district), Panel (b) shows the linear fit of those regression coefficients, the location exposure effect. The regression in Panel (a) controls for birth cohort fixed effects and linearly for origin quality interacted with birth cohort fixed effects (Equation 5). Maximum age at migration is 20 years. Sample: one-time migrants age 21+ from SUPAS1985, SUPAS1995, SUPAS2005. Figure 4: Location Exposure Effect on Years of Education (Adults Age 21+) Selection into Migration R Note: The figure shows the effect of location quality change ∆Y odc on years of education by age at migration during childhood, and the linear fit of those regression coefficients, the location exposure effect. Linear fit estimated separately for migrations until the age of 20, and between 21 and 40. The regression coefficients are obtained controlling for birth cohort fixed effects and linearly for origin quality interacted with birth cohort fixed effects (Equation 5). Maximum age at migration is 40 years. Sample: one-time migrants age 21+ from SUPAS1985, SUPAS1995, SUPAS2005. Tables Table 1: Migration and Education Dependent Variable: Primary (1) Junior Secondary (2) (3) Senior Secondary (4) GFA (5) (6) (7) (8) (a) Population Census 2000 0.019∗∗∗ 0.001 0.069∗∗∗ 0.011∗∗∗ 0.090∗∗∗ 0.018∗∗∗ 0.062∗∗∗ 0.012∗∗∗ (0.002) (0.001) (0.005) (0.002) (0.005) (0.002) (0.004) (0.002) Parental Education No Yes No Yes No Yes No Yes Dep. Var. Mean 0.82 0.82 0.56 0.56 0.32 0.32 0.49 0.49 29,853,484 29,853,484 19,231,384 19,231,384 8,802,663 8,802,663 29,853,484 29,853,484 Migrated Observations (b) Population Census 2010 0.023∗∗∗ 0.002∗∗ 0.076∗∗∗ 0.011∗∗∗ 0.122∗∗∗ 0.021∗∗∗ 0.076∗∗∗ 0.013∗∗∗ (0.001) (0.001) (0.006) (0.002) (0.008) (0.003) (0.005) (0.002) Parental Education No Yes No Yes No Yes No Yes Dep. Var. Mean 0.86 0.86 0.70 0.70 0.46 0.46 0.62 0.62 29,556,509 29,556,509 18,234,321 18,234,321 8,108,623 8,108,623 29,556,509 29,556,509 Migrated Observations Note: ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. Standard errors clustered (two-way) at origin and destination district level. Outcome variables are binary indicators for having completed at least elementary school, junior high school, or senior high school, respectively. GFA indicates the completion of the appropriate degree given the child’s age. Migration defined as birth district different to current district. Controls for origin and destination district fixed effects, and birth year fixed effects. Even columns additionally control flexibly for the education of the household head and their spouse, interacted with the gender of head and spouse, respectively. Columns 1, 2, 7, and 8 are estimated on the sample of 12 - 20 year olds cohabitating with at least one parent in the 2000 and 2010 Census, respectively. The sample in Columns 3 and 4 is restricted to 15- to 20-year-olds, in Column 5 and 6 to 18- to 20-year-olds. Sample: entire population age 12 - 20, from 2000 and 2010 Indonesian Population Census. Table 2: Destination Quality and Education Dependent Variable: Primary12+ JUNSEC15+ SENSEC18+ GFA (1) (2) (3) (4) 0.675∗∗∗ primary pdc (0.022) 0.603∗∗∗ JU N SEC pdc (0.023) 0.607∗∗∗ SEN SEC pdc (0.026) 0.721∗∗∗ GF Apdc (0.020) Dep. Var. Mean Observations Note: ∗ p < 0.05, ∗∗ 0.88 0.74 0.54 0.66 5,964,313 3,847,739 1,788,362 5,964,313 p < 0.01, ∗∗∗ p < 0.001. Standard errors clustered two-way on origin district (in parentheses). Controls for fixed effects of origin and destination district, birth cohort, and education of household head interacted with gender of household head. Sample: one-time migrants age 12 - 20 from 2000 and 2010 Population Census. Table 3: Location Exposure Effect on Graduation Rates Dependent Variable: Primary12+ JUNSEC15+ SENSEC18+ GFA (1) (2) (3) (4) -0.019∗∗∗ -0.018∗∗∗ -0.018∗∗∗ -0.019∗∗∗ (0.003) (0.002) (0.002) (0.002) Sample Mean 0.86 0.73 0.53 0.75 Observations 1,418,423 874,279 393,669 1,455,965 11 14 17 17 Age Migrated × c.Delta Max. Mig. Age Note: ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. Standard errors clustered two-way on origin and destination district (in parentheses). Reports Reports location exposure effect on primary school completion (Primary12+) for children age 12 to 20, junior secondary school completion (JUNSEC15+) for children age 15 - 20, senior secondary school lin from Equation completion (SENSEC18+) for children age 18 - 20, and grade-for-age (GFA) for children age 12 - 20 (β̂m 6). Controls for origin × birth-cohorts fixed effects, age-at-migration × parent-education group fixed effects, and highest completed degree of the household head interacted with gender of household head, and highest completed degree of their spouse interacted with the spouse’s gender. Max. mig. age reports the sample’s respective maximum age of migration. Sample: one-time migrants from 2000 and 2010 Indonesian Population Census. Table 4: Heterogeneity Location Exposure Effect on Junior Secondary School Graduation Sample Restriction: Low Parental High Parental Baseline Education Education Better Worse (1) (2) (3) (4) (5) -0.018∗∗∗ -0.017∗∗∗ -0.009∗∗ -0.016∗∗∗ -0.018∗∗∗ (0.002) (0.002) (0.003) (0.003) (0.004) Sample Mean 0.73 0.53 0.86 0.74 0.72 Observations 874,279 338,792 535,368 402,096 471,918 N SEC Age Migrated × ∆JU podc Note: ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. Standard errors clustered two-way on origin and destination district (in parentheses). lin from Equation 6). Reports location exposure effect on junior secondary school completion (JUNSEC15+) for children age 15 - 20 (β̂m Controls for origin × birth-cohorts fixed effects, age-at-migration × parent-education group fixed effects, and highest completed degree of the household head interacted with gender of household head, and highest completed degree of their spouse interacted with the spouse’s gender. Maximum age at migration is 14 years. Columns 1 shows full sample, Column 2 restricts to children with parents that have not completed junior secondary school, Column 3 restricts to children with at least one parent that has at least junior secondary N SEC > 0, and Column 5 restricts to moves with ∆JU N SEC < 0. Sample: school completed, Column 4 restricts to moves with ∆JU podc podc one-time migrants from 2000 and 2010 Indonesian Population Census. Table 5: Location Exposure Effect on Completed Education And Labor Market Outcomes Dependent Variable: Years of Schooling Formal Employment Log Income (1) (2) (3) -0.007∗ -0.001∗∗ -0.001 (0.003) (0.000) (0.001) Sample Mean 8.76 0.47 12.01 Observations 50,254 30,092 7,131 R Age Migrated × ∆Ypodc Note: ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. Standard errors clustered two-way on origin and destination district (in lin from Equation 6) with district quality change measured in average parentheses). Reports location exposure effect (β̂m years of education of permanent residents. Controls for birth cohort fixed effects and linearly for origin quality interacted with birth cohort fixed effects. Maximum age at migration is 20 years. The dependent variable in Column (1) is years of schooling, in Column (2) an indicator for employment in the formal labor market, and in Column (3) log income. Sample: one-time migrants age 21+ from SUPAS1985, SUPAS1995, SUPAS2005. Table 6: Robustness Location Exposure Effect on Junior Secondary School Graduation: Selection Specification: Baseline No Parent Controls Household FE Baseline FE Sample (1) (2) (3) (4) -0.018∗∗∗ -0.027∗∗∗ -0.028∗∗∗ -0.020∗∗∗ (0.002) (0.003) (0.008) (0.003) Sample Mean 0.73 0.73 0.71 0.71 Observations 874,279 874,279 247,957 247,957 N SEC Age Migrated × ∆JU podc Note: ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. Standard errors clustered two-way on origin and destination district (in parentheses). Reports location lin from Equation 6). Baseline regression in Column exposure effect on junior secondary school completion (JUNSEC15+) for children age 15 - 20 (β̂m (1) controls for origin × birth-cohorts fixed effects, age-at-migration × parent-education group fixed effects, and highest completed degree of the household head interacted with gender of household head, and highest completed degree of their spouse interacted with the spouse’s gender. Column (2) omits all parent related controls, Column (3) additionally controls for household fixed effects, and Column (4) repeats the baseline analysis on the sample from Column (3). Maximum age at migration is 14 years. Sample: one-time migrants from 2000 and 2010 Indonesian Population Census. Table 7: Robustness Location Exposure Effect on Primary School Completion: Age of Child Sample Restriction: Baseline Age 12 - 18 Age 12 - 16 Age 12 - 14 (1) (2) (3) (4) -0.019∗∗∗ -0.018∗∗∗ -0.019∗∗∗ -0.017∗∗∗ (0.003) (0.003) (0.004) (0.004) Sample Mean 0.86 0.85 0.81 0.73 Observations 1,418,423 1,184,423 892,658 571,042 Age Migrated × ∆primary podc Note: ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. Standard errors clustered two-way on origin and destination district lin from Equation 6). Controls (in parentheses). Reports location exposure effect on primary school completion (β̂m for origin × birth-cohorts fixed effects, age-at-migration × parent-education group fixed effects, and highest completed degree of the household head interacted with gender of household head, and highest completed degree of their spouse interacted with the spouse’s gender. Maximum age at migration is 11 years. Column (1) reproduces baseline regression with the outcome measured for children age 12 - 20, Column (2) restricts the age of the child to 12 - 18 years, Column (3) restricts to 12 - 16 years, Column (4) to 12 - 14. Sample: one-time migrants from 2000 and 2010 Indonesian Population Census. Appendix Table A.1: Robustness Destination Effect on Junior Secondary School Graduation: Age of Child Sample Restriction: Age 15 - 20 Age 15 - 19 Age 15 - 18 Age 15 - 17 (1) (2) (3) (4) -0.018∗∗∗ -0.018∗∗∗ -0.018∗∗∗ -0.017∗∗∗ (0.002) (0.002) (0.002) (0.002) Sample Mean 0.73 0.72 0.70 0.68 Observations 874,279 753,458 630,602 487,394 N SEC Age Migrated × ∆JU podc Note: ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. Standard errors clustered two-way on origin and destination district (in lin from Equation 6). Controls parentheses). Reports location exposure effect on junior secondary school completion (β̂m for origin × birth-cohorts fixed effects, age-at-migration × parent-education group fixed effects, and highest completed degree of the household head interacted with gender of household head, and highest completed degree of their spouse interacted with the spouse’s gender. Maximum age at migration is 14 years. Column (1) reproduces baseline regression with the outcome measured for children age 15 - 20, Column (2) restricts the age of the child to 15 - 19 years, Column (3) restricts to 15 - 18 years, Column (4) to 15 - 17. Sample: one-time migrants from 2000 and 2010 Indonesian Population Census. Table A.2: Robustness Location Exposure Effect on Junior Secondary School Graduation: Year of Migration Estimate Sample Restriction: Error ≤ 4 Error ≤ 3 Error ≤ 2 Error ≤ 1 (1) (2) (3) (4) -0.018∗∗∗ -0.018∗∗∗ -0.017∗∗∗ -0.017∗∗∗ (0.002) (0.002) (0.002) (0.003) Sample Mean 0.73 0.72 0.72 0.71 Observations 874,279 712,313 481,925 202,850 N SEC Age Migrated × ∆JU podc Note: ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. Standard errors clustered two-way on origin and destination lin from district (in parentheses). Reports location exposure effect on junior secondary school completion (β̂m Equation 6). Controls for origin × birth-cohorts fixed effects, age-at-migration × parent-education group fixed effects, and highest completed degree of the household head interacted with gender of household head, and highest completed degree of their spouse interacted with the spouse’s gender. Maximum age at migration is 14 years. Column (1) reproduces baseline regression with a maximum error of 4 years, Column (2) restricts to maximum year of migration estimate error of 3 years, Column (3) restricts to maximim error of 2 years, Column (4) to 1 year. Sample: one-time migrants from 2000 and 2010 Indonesian Population Census. Table A.3: Robustness Destination Effect on Junior Secondary School Graduation: Number of Children in Household, Inter Province Migration Baseline Sample Restriction: Number of Number of Migration Across Children Children ≥4 ≤3 Provinces Migration Within Provinces (1) (2) (3) (4) (5) -0.018∗∗∗ -0.018∗∗∗ -0.017∗∗∗ -0.019∗∗∗ -0.016∗∗∗ (0.002) (0.002) (0.002) (0.002) (0.003) Sample Mean 0.73 0.68 0.77 0.73 0.73 Observations 874,279 391,090 483,078 474,453 399,669 N SEC Age Migrated × ∆JU podc Note: ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. Standard errors clustered two-way on origin and destination district (in parentheses). lin from Equation 6). Controls for origin × birth-cohorts Reports location exposure effect on junior secondary school completion (β̂m fixed effects, age-at-migration × parent-education group fixed effects, and highest completed degree of the household head interacted with gender of household head, and highest completed degree of their spouse interacted with the spouse’s gender. Maximum age at migration is 14 years. Column (1) reproduces baseline regression on the full sample. Column (2) keeps families with at least fourchildren, Column (3) restricts to families with two or three children, Column (4) restricts to migration across province borders, Column (5) restricts to migration within the same province. Table A.4: Robustness Destination Effect on Junior Secondary School Graduation: Population Size Sample Restriction: Pop ≥ 30 Pop ≥ 100 Pop ≥ 200 Pop ≥ 300 (1) (2) (3) (4) -0.018∗∗∗ -0.018∗∗∗ -0.018∗∗∗ -0.018∗∗∗ (0.002) (0.002) (0.002) (0.002) Sample Mean 0.73 0.73 0.73 0.73 Observations 874,279 873,458 869,140 860,386 N SEC Age Migrated × ∆JU podc Note: ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. Standard errors clustered two-way on origin and destination lin from district (in parentheses). Reports location exposure effect on junior secondary school completion (β̂m Equation 6). Controls for origin × birth-cohorts fixed effects, age-at-migration × parent-education group fixed effects, and highest completed degree of the household head interacted with gender of household head, and highest completed degree of their spouse interacted with the spouse’s gender. Maximum age at migration is 14 N SEC based on at least 30 permanent resident chilyears. Column (1) reproduces baseline regression with ∆JU podc dren in each birth cohort × parent education group × district cell, Column (2) restricts to at least 100 permanent resident children in each cell, Column (3) restricts to at least 200 children in each cell, Column (4) to 300 children. Sample: one-time migrants from 2000 and 2010 Indonesian Population Census.