Family Wealth and Children's Educational Outcomes in Sweden - an Extended Family Approach

We study the role of family wealth for children’s educational outcomes using unique Swedish population level register data for cohorts born in 1987 to 1996. The data identifies different components of wealth, where a crucial distinction is between liquid (cash and financial assets) and real wealth (houses, items), and debt. We measure family wealth in three dimensions: parents’ wealth, maternal grandparents’ wealth, and aunts/uncles wealth. Our models use fixed effects for paternal grandparents in order control for unobserved heterogeneity, we thus compare cousins which differ in their maternal lineage. We also control for other SES factors such as occupation and education (and to some extent also income) in each dimension, as well as cognitive ability and non-cognitive skills of fathers and uncles using military enlistment data, and contextual SES factors of children’s’ schools. Net of all social background factors, our preliminary results suggest that that aunts/uncles wealth plays no substantial role for 9th grade GPA in the non-tracked Swedish elementary school. We find small to intermediate substantial effects of parents and grandparents wealth, and it is liquid wealth that plays the most important role. For example, comparing children with zero grandparental liquid wealth with those in the 95^th percentile, yield a difference of 1/15 of a standard deviation (SD) in GPA. With the same comparison for grandparent’s real wealth, the effect is 1/20 of an SD, and there is a similarly sized negative effect of grandparental debt. For parents’ liquid wealth, the effect is 1/6 of a SD in GPA, and 1/10 of a SD in GPA. We find similarly sized effect effects on subsequent outcomes, e.g., transition to upper secondary school, track choice in upper-secondary school, and upper-secondary school GPA, but 9^th grade GPA mediates most of these effects. We conclude that the effect of wealth in Sweden is concentrated to educational achievement, not educational choices, and we discuss the possible mechanisms which may explain this finding.