The Impact of Family Composition on Educational Achievement

Note: This table reports the OLS estimated effect of sibling gender composition on family size. Other covariates included in Columns 1–3 are the full set of dummies for urban, the subject’s age, parents’ years of birth, and maternal age at the first birth. Columns 4–5 split the sample by the urban dummy and Columns 6–9 by the indicator for mother’s or father’s high school diploma. The top panel reports the estimates for all families whose firstborn singleton was born between 1978 and 1984, and the second part of the table reports the estimates for the same set of firstborns but restricted to those having at least one sibling. Robust standard errors are in parentheses.

↵* indicates significance at the 5% level. We assume in Model II that the coefficients of Boy1st and Boy2nd are approximately equal, as confirmed by the result in Model I.

Notes: This table reports descriptive statistics of family characteristics of firstborn singletons who were born between 1978 and 1984 and had at least one sibling. Columns 2 and 6 include those whose next sibling was born before 1985, and Columns 3 and 7 include those whose next sibling was born afterwards. We exclude families with father or mother younger than 18 when their first child was born, or with missing information about child birth year, or with the number of children at some parity exceeding three.

↵* indicates significance at the 5% level.

Notes: This table reports the OLS estimated coefficients for a regression of birth spacing between the first two births on their sex composition. We use the same sample as the regressions in Tables 6 to 11, except for 811 families whose second born children have missing or erroneous birthday information. We restrict the sample in Column 3 to those whose first child was born from 1980 onwards the sample in Column 4 to those whose first child was born from 1980 onwards and with the second birth prior to 1985. Additional covariates include the subject’s age and district of birth, indicators for urban, parents’ education and years of birth, mother’s age at first birth. Robust standard errors are reported in parentheses.

↵* indicates significance at the 5% level.

Notes: The table reports the OLS estimated coefficients of Boy2nd in a regression of urban or each parental education level. Covariates not listed here are the full set of indicators for the district of firstborn’s birthplace, parents’ years of birth, and maternal age at the first birth.

↵* indicates the 5% significant level.

Notes: This table reports the OLS results of how the firstborn’s education is associated with the sex of the secondborn (D), having more than two children (M), and their interaction. We also report in Appendix Table A5 the corresponding results with M being measured by sibsize. We include 416,315 firstborn females and 434,729 firstborn males born between 1978 and 1984 who have at least one sibling. We control for the full set of dummies for urban, the subject’s age and district of birth, parents’ education and years of birth, and mother’s age at the first birth. The sample mean of the educational outcome (Y) is in italics, and robust standard errors are in parentheses.

Notes: Panels 1 and 3 of this table report the first-stage estimates for fertility choice, using twinning at the second birth (Twin2nd) as instrument. Panels 2 and 4 report the first-stage estimates for the interaction between a next brother D and fertility choice, using D × Twin2nd as instrument. An interaction term for D × Twin2nd is included in Columns 2 and 4 when the interaction between D and fertility choice is included in the outcome equation. We include 416,315 firstborn females and 434,729 first born males born between 1978 and 1984 who have at least one sibling. Additional covariates include the full set of indicators for urban, the subject’s age and district of birth, parents’ education and years of birth, and mother’s age at the first birth. The multivariate first-stage F-statistics are constructed as described in Angrist and Pischke (2009, p. 217–218). Means of the fertility choice variable are in italics, and robust standard errors are in parentheses.

Notes: This table reports OLS and IV estimates for the educational outcome equation (with university admission as dependent variable) and decomposed effects of sibling gender on the outcome, with fertility choice being measured by Morethan2. The corresponding results for fertility choice being measured by Sibsize are reported in Table 11. The first-stage F-statistics are reported in Table 7. The samples include 416,315 firstborn females and 434,729 firstborn males, who were born between 1978 and 1984 and have at least one sibling. We exclude the interaction between the twins instrument and sibling gender from the first stage for Columns 2 and 7, while we include it for Columns 3 and 8 in order to be comparable to Columns 5 and 10. The decomposed effects are measured according to Equations 4–6 as summarized: AIE = β₂{E[M|D = 1] – E[M|D = 0]}, ADE = β₁ + β₃E[M|D = 1], CDE = β₁ + β₃E[M]. AIE and ADE are evaluated at the conditional mean of M=Morethan2: (E[M|D = 1], E[M|D = 0]) = (0.4787, 0.6992) for firstborn females and (0.4191, 0.4831) for firstborn males, while CDE is evaluated at the unconditional mean E[M] = 0.5854 for firstborn females and 0.4503 for firstborn males. We control for the full set of dummies for urban, the subject’s age and district of birth, parents’ education and years of birth, and mother’s age at the first birth. Robust standard errors are reported in parentheses.

Notes: This table reports OLS and IV estimates for the educational outcome equation (with university admission as dependent variable) and decomposed effects of sibling gender on the outcome, with fertility choice being measured by Morethan2. The corresponding results for fertility choice being measured by Sibsize are reported in Table 11. The first-stage F-statistics are reported in Table 7. The samples include 416,315 firstborn females and 434,729 firstborn males, who were born between 1978 and 1984 and have at least one sibling. We exclude the interaction between the twins instrument and sibling gender from the first stage for Columns 2 and 7, while we include it for Columns 3 and 8 in order to be comparable to Columns 5 and 10. The decomposed effects are measured according to Equations 4–6 as summarized: AIE = β₂{E[M|D = 1] – E[M|D = 0]}, ADE = β₁ + β₃E[M|D = 1], CDE = β₁ + β₃E[M]. AIE and ADE are evaluated at the conditional mean of M=Morethan2: (E[M|D = 1], E[M|D = 0]) = (0.4787, 0.6992) for firstborn females and (0.4191, 0.4831) for firstborn males, while CDE is evaluated at the unconditional mean E[M] = 0.5854 for firstborn females and 0.4503 for firstborn males. We control for the full set of dummies for urban, the subject’s age and district of birth, parents’ education and years of birth, and mother’s age at the first birth. Robust standard errors are reported in parentheses.

Notes: This table reports OLS and IV estimates for the educational outcome equation (with high school admission as dependent variable) and decomposed effects of sibling gender on the outcome, with fertility choice being measured by Sibsize. The corresponding results for fertility choice being measured by Morethan2 are reported in Table 8. The first-stage F-statistics are reported in Table 7. The samples include 416,315 firstborn females and 434,729 firstborn males, who were born between 1978 and 1984 and have at least one sibling. We exclude the interaction between the twins instrument and sibling gender from the first stage for Columns 2 and 7, while we include it for Columns 3 and 8 in order to be comparable to Columns 5 and 10. The decomposed effects are measured according to Equations 4–6 as summarized: AIE = β₂{E[M|D = 1] – E[M|D = 0]}, ADE = β₁ + β₃E[M|D = 1], CDE = β₁ + β₃E[M]. AIE and ADE are evaluated at the conditional mean of M = Sibsize: (E[M|D = 1], E[M|D = 0]) = (2.5947, 3.0255) for first born females and (2.4975, 2.5993) for first born males, while CDE is evaluated at the unconditional mean E[M] = 2.8032 for firstborn females and 2.5471 for firstborn males. We control for the full set of dummies for urban, the subject’s age and district of birth, parents’ education and years of birth, and mother’s age at the first birth. Robust standard errors are reported in parentheses.

Notes: This table reports OLS and IV estimates for the educational outcome equation (with university admission as dependent variable) and decomposed effects of sibling gender on the outcome, with fertility choice being measured by Sibsize. The corresponding results for fertility choice being measured by Morethan2 are reported in Table 9. The first-stage F-statistics are reported in Table 7. The samples include 416,315 firstborn females and 434,729 firstborn males, who were born between 1978 and 1984 and have at least one sibling. We exclude the interaction between the twins instrument and sibling gender from the first stage for Columns 2 and 7, while we include it for Columns 3 and 8 in order to be comparable to Columns 5 and 10. The decomposed effects are measured according to Equations 4–6 as summarized: AIE = β₂{E[M|D = 1] – E[M|D = 0]}, ADE = β₁ + β₃E[M|D = 1], CDE = β₁ + β₃E[M]. AIE and ADE are evaluated at the conditional mean of M = Sibsize: (E[M|D = 1], E[M|D = 0]) = (2.5947, 3.0255) for first born females and (2.4975, 2.5993) for first born males, while CDE is evaluated at the unconditional mean E[M] = 2.8032 for firstborn females and 2.5471 for firstborn males. We control for the full set of dummies for urban, the subject’s age and district of birth, parents’ education and years of birth, and mother’s age at the first birth. Robust standard errors are reported in parentheses.

Notes: This table reports robustness checks on the 2SLS results for the educational outcome equation and the decomposed effects of sibling gender on firstborn females. Though not reported here, we find no evidence of nonlinearity in gestation length. Since boys are heavier than girls at birth on average, we use gender-specific birthweight percentiles and calculate the mean percentile of secondborn twins. We control implicitly for the full set of dummies for urban, the subject’s age and district of birth, parents’ education and years of birth, and maternal age at the first birth. Robust standard errors are reported in parentheses. Estimates for firstborn males are reported in Appendix Table A7.

Notes: This table reports the OLS estimated coefficient of demographic covariates in the regression D or Twin2nd by firstborn gender, conditional on the full set of indicators for urban, the firstborn’s age and district of birthplace, parents’ years of birth, and maternal age at the first birth.

↵* indicates significance at the 5% level.