Working Paper: CEPR ID: DP3354
Authors: Dorothe Bonjour; Lynn Cherkas; Jonathan Haskel; Denise Hawkes; Tim Spector
Abstract: We use a new sample of UK female identical twins to estimate private economic returns to education. We report findings in three areas. First, we use identical twins, to control for family effects and genetic ability bias, and the education reported by the other twin to control for schooling measurement error. Our estimates suggest a return to schooling for UK females of about 7.7%. Second, we investigate within-twin pair ability differences by examining within-twin pair and between-family correlations of education with observable correlates of ability (including birthweight, ability tests and reading scores). Our findings suggest lower ability bias in within-twin pair regressions than pooled regressions. Third, using data on twins smoking we show smoking reflects family background and using it as an instrument exacerbates ability bias.
Keywords: ability bias; measurement error; returns to education; smoking; twins
JEL Codes: I21; J24; J31
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| education reported by one twin (I24) | education reported by the other twin (I24) |
| education (I29) | return to schooling (instrumented) (I26) |
| within-twin pair estimates (C21) | return to schooling (I26) |
| ability bias (D91) | return to schooling (pooled estimates) (I26) |
| measurement error (C20) | return to schooling (I26) |
| education (I29) | return to schooling (I26) |