This paper discusses the problem of errors in variables in panel data, which is a common issue in econometric analysis. Panel data, derived from longitudinal surveys, allows for the control of individual effects and other slowly changing variables. However, the "within" results from such data are often unsatisfactory, "too low" and insignificant. This is often attributed to measurement errors in the independent variables, which are magnified in the within dimension. The paper argues that the standard errors-in-variables model has not been widely applied in panel data contexts because it requires extraneous information to identify the parameters of interest. However, in the panel data context, various errors-in-variables models can be identified and estimated without the use of external instruments. The authors develop this idea and illustrate its application in a relatively simple but not uninteresting case: the estimation of "labor demand" relationships, also known as the "short run increasing returns to scale" puzzle. The paper presents a detailed analysis of the first difference and within estimators, showing that the first difference estimator is more prone to bias due to measurement errors. The authors also discuss alternative estimation strategies, including the use of "long" differences, which can provide more accurate estimates of the parameters. The paper concludes that the use of instrumental variable methods and the consideration of measurement errors in the model can lead to more accurate estimates of the parameters of interest. The authors also present an empirical example using data on U.S. manufacturing firms, showing that the estimated labor demand relationships are consistent with the predictions of the errors-in-variables model. The results suggest that the true value of the labor demand elasticity is around 0.78, and that the unanticipated variance in output accounts for a significant portion of the observed variance in the first differences. The paper concludes that the errors-in-variables model is a useful tool for analyzing panel data and that the use of appropriate estimation techniques can lead to more accurate estimates of the parameters of interest.This paper discusses the problem of errors in variables in panel data, which is a common issue in econometric analysis. Panel data, derived from longitudinal surveys, allows for the control of individual effects and other slowly changing variables. However, the "within" results from such data are often unsatisfactory, "too low" and insignificant. This is often attributed to measurement errors in the independent variables, which are magnified in the within dimension. The paper argues that the standard errors-in-variables model has not been widely applied in panel data contexts because it requires extraneous information to identify the parameters of interest. However, in the panel data context, various errors-in-variables models can be identified and estimated without the use of external instruments. The authors develop this idea and illustrate its application in a relatively simple but not uninteresting case: the estimation of "labor demand" relationships, also known as the "short run increasing returns to scale" puzzle. The paper presents a detailed analysis of the first difference and within estimators, showing that the first difference estimator is more prone to bias due to measurement errors. The authors also discuss alternative estimation strategies, including the use of "long" differences, which can provide more accurate estimates of the parameters. The paper concludes that the use of instrumental variable methods and the consideration of measurement errors in the model can lead to more accurate estimates of the parameters of interest. The authors also present an empirical example using data on U.S. manufacturing firms, showing that the estimated labor demand relationships are consistent with the predictions of the errors-in-variables model. The results suggest that the true value of the labor demand elasticity is around 0.78, and that the unanticipated variance in output accounts for a significant portion of the observed variance in the first differences. The paper concludes that the errors-in-variables model is a useful tool for analyzing panel data and that the use of appropriate estimation techniques can lead to more accurate estimates of the parameters of interest.