Finite Sample Bias In Iv Estimation Of Intertemporal Labor Supply Models: Is The Intertemporal Substitution Elasticity Really Small?

The empirical literature on intertemporal labor supply behavior documents that the intertemporal elasticity of substitution of labor supply is very low, with a plausible range of zero to 0.2. Drawing upon the literature on the distribution of instrumental variables (IV) estimators, this paper demonstrates that this conventional wisdom is erroneous because it does not take into account the severe finite sample bias in these estimates that arise from weak instruments. This paper adopts several approaches to adjust for the problems induced by these weak instruments. The empirical results show that, when uncorrected for finite sample bias, the two-stage least-squares (2SLS) estimate of the elasticity is essentially zero, as in most of the previous studies, with its valid confidence interval being open-ended, [-∞, +∞]. However, when corrected for finite sample bias, the estimate becomes approximately 0.5 with a much tighter confidence interval. © 2001 by the President and Fellows of Harvard College and the Massachusetts Institute of Technology