| Problems with measurement error and possible wealth effects that accompany wage changes have led many researchers to criticize empirical studies of the intertemporal substitution hypothesis that use panel data. In this study I address the measurement error and wealth effect problems explicitly. I use data from the Panel Study of Income Dynamics Validation Study to estimate the properties of measurement error in the Panel Study of Income Dynamics. I show there exists a large transitory component to wages, even after accounting for measurement error. I then show that while wage changes may not be predictable, transitory wages predictably disappear. When estimating the labor supply response to these predictable wage changes, I account for serially correlated measurement error and for measurement error that is correlated with true hours and wages. I show that failure to control for serially correlated measurement error that is correlated with true variables leads to an upward bias in estimates of the intertemporal elasticity of substitution. I find that the intertemporal elasticity of substitution is -0.4 when controlling for the correlation properties of measurement error. However, the estimate is .3 if the correlation properties of measurement error ignored, as is the case in other studies. In other words, I find evidence against the auction model of the labor market. I also find a reason why many studies could have estimates to the contrary. |