This paper by Zvi Griliches and Jerry A. Hausman explores the issue of errors in variables in panel data, a common problem in economics where longitudinal surveys are used. The authors argue that the standard errors-in-variables model, which often requires extraneous information to identify parameters, can be applied more widely in panel data contexts without external instruments. They develop and illustrate this idea through a simple but significant example: estimating "labor demand" relationships, also known as the "short-run increasing returns to scale" puzzle. The paper outlines an approach to estimating the true parameters by comparing the biases introduced by errors in measurement in the "within" and "first difference" estimators. It also provides a method for consistent estimation using instrumental variables and discusses the implications for the "short-run increasing returns to scale" puzzle. The empirical example uses data from 1242 U.S. manufacturing firms over six years (1972-1977) to illustrate the methods and concludes that the true elasticity of labor demand with respect to output is likely between 0.75 and 0.85, rather than the commonly accepted value of 1.This paper by Zvi Griliches and Jerry A. Hausman explores the issue of errors in variables in panel data, a common problem in economics where longitudinal surveys are used. The authors argue that the standard errors-in-variables model, which often requires extraneous information to identify parameters, can be applied more widely in panel data contexts without external instruments. They develop and illustrate this idea through a simple but significant example: estimating "labor demand" relationships, also known as the "short-run increasing returns to scale" puzzle. The paper outlines an approach to estimating the true parameters by comparing the biases introduced by errors in measurement in the "within" and "first difference" estimators. It also provides a method for consistent estimation using instrumental variables and discusses the implications for the "short-run increasing returns to scale" puzzle. The empirical example uses data from 1242 U.S. manufacturing firms over six years (1972-1977) to illustrate the methods and concludes that the true elasticity of labor demand with respect to output is likely between 0.75 and 0.85, rather than the commonly accepted value of 1.