Double trouble: on the value of twins-based estimation of the return to schooling

Several recent studies, including the Rouse (1999) and Behrman & Rosenzweig (1999) articles in this issue, use the schooling and wage variation betwee n monozygotic twins to estimate the return to schooling. In this overview a rticle, we summarize the results from this literature, and we examine the i mplications of endogenous determination of which twin goes to school longer and of measuring schooling with (possibly mean-reverting) error. Endogenei ty of between-twins schooling variation is strongly suggested by the extens ive (mostly non-economic) literature documenting that the between-twins dif ference in birth weight is correlated with the between-twins differences in both schooling and IQ. We conclude that twins-based estimation is vulnerab le to the same sort of inconsistency that afflicts conventional cross-secti onal estimation. We argue, however, that, if one starts with the presumptio n that endogenous schooling induces upward inconsistency in the estimated r eturn to schooling, the new twins-based estimates may complement other appr oaches to tightening the upper bound on the return to schooling. (C) 1999 E lsevier Science Ltd. All rights reserved.