The authors of this NBER working paper analyze the relationship between current and lifetime earnings, finding that the textbook errors-in-variables model does not accurately describe this relationship. Using data from the Health and Retirement Study, they show that the association between current and lifetime earnings varies systematically over the life cycle. This variation has important implications for errors-in-variables bias in econometric research that uses current earnings as a proxy for lifetime earnings. The authors develop models to illustrate how current earnings vary with lifetime earnings over the life cycle. They find that the slope coefficient in the regression of current earnings on lifetime earnings is not generally equal to 1, but instead varies over the life cycle. This finding challenges the traditional assumption that current earnings are a perfect proxy for lifetime earnings. The authors also show that errors-in-variables bias can be either an attenuation bias or an amplification bias, depending on the age at which earnings are measured. They find that using current earnings in the early stages of a career tends to underestimate the relationship between current and lifetime earnings, while using current earnings in later stages tends to overestimate it. The authors use a detailed dataset of Social Security earnings histories to estimate the association between current and lifetime earnings. They find that the correlation between current and lifetime earnings is strong in later life but weak in early life. They also find that the reliability ratio, which measures the accuracy of current earnings as a proxy for lifetime earnings, varies over the life cycle. The authors conclude that the traditional errors-in-variables model is not appropriate for analyzing the relationship between current and lifetime earnings. Instead, they recommend using models that account for the life-cycle variation in this relationship. Their findings have important implications for a wide range of research that uses current earnings as a proxy for lifetime earnings.The authors of this NBER working paper analyze the relationship between current and lifetime earnings, finding that the textbook errors-in-variables model does not accurately describe this relationship. Using data from the Health and Retirement Study, they show that the association between current and lifetime earnings varies systematically over the life cycle. This variation has important implications for errors-in-variables bias in econometric research that uses current earnings as a proxy for lifetime earnings. The authors develop models to illustrate how current earnings vary with lifetime earnings over the life cycle. They find that the slope coefficient in the regression of current earnings on lifetime earnings is not generally equal to 1, but instead varies over the life cycle. This finding challenges the traditional assumption that current earnings are a perfect proxy for lifetime earnings. The authors also show that errors-in-variables bias can be either an attenuation bias or an amplification bias, depending on the age at which earnings are measured. They find that using current earnings in the early stages of a career tends to underestimate the relationship between current and lifetime earnings, while using current earnings in later stages tends to overestimate it. The authors use a detailed dataset of Social Security earnings histories to estimate the association between current and lifetime earnings. They find that the correlation between current and lifetime earnings is strong in later life but weak in early life. They also find that the reliability ratio, which measures the accuracy of current earnings as a proxy for lifetime earnings, varies over the life cycle. The authors conclude that the traditional errors-in-variables model is not appropriate for analyzing the relationship between current and lifetime earnings. Instead, they recommend using models that account for the life-cycle variation in this relationship. Their findings have important implications for a wide range of research that uses current earnings as a proxy for lifetime earnings.