Recent within-twin estimates of the return to schooling that use instrument
al variables to correct for measurement error are considerably higher than
existing estimates. This paper extends Griliches' [Griliches (1979) Sibling
models and data in economics: beginnings of a survey, Journal of Political
Economy 87, S37-S64]. analysis of sibling or twin estimates of the return
to schooling to show that small ability differences among twins can lead to
more upward omitted ability bias-and more upward bias overall-in the instr
umental variables estimate that corrects for measurement error than in the
standard within-twin estimate. This point is relevant for any application i
n which differencing may not fully remove the omitted variable, and may exp
lain high within-twin instrumental variable estimates of the return to scho
oling. [JELI21] (C) 1999 Elsevier Science Ltd. All rights reserved.