matched data. These studies also show a mixed pattern, with many studies estimating peak productivity at around 40 years of age, while a few find peak productivity at older ages. Case studies in the United States and Germany shed more light on the age-productivity relation. Kotlikoff and Wise (1989) found that the productivity of salespeople in a large insurance company, measured by the value of contracts sold, increases with age. Boersch-Supan, Duezguen, and Weiss (2008) and Boersch-Supan and Weiss (2011) assembled a large data set on production workers in a German car manufacturing company over many years and show a similar effect. They measure productivity by the absence of errors in a well-defined production process. They find that, while the number of small errors is larger among older workers, major errors are more frequent among younger ones. Their measure of productivity finds that older workers have higher productivity.

Although the literature on productivity and behavior at the individual level provides weighty evidence on the impact of aging on many individual attributes, we need to be cautious about the application of those attributes to aggregate productivity. Many of the studies are cross-sectional and do not take into account changes in occupation or, in labor market studies, attrition.

Additionally, the determinants of individual productivity are extremely complex and are unlikely to be captured in most metrics. For example, a typical cross-sectional study of earnings can explain a small fraction of the dispersion on the basis of personal attributes such as intelligence. Moreover, while it is true in a few areas that reasonable output metrics have been developed (such as for athletes), we know that in other areas the measures have often proven highly unreliable and even systemically dangerous (such as the compensation metrics used in many financial firms).

Furthermore, the important skill sets, and the difficulties in accurately measuring them, will change over the life cycle as workers move from being unskilled workers at fast food stores in summer jobs, to entry-level technicians, to middle and upper management positions. Given the multitude of attributes and vast number of different jobs, it has proven very difficult to make an accurate measure of the economic value of an individual’s attributes and the changes in those over time due to aging.

A final reason to discount metrics on individual attributes is that work increasingly takes place in teams. Teams are often composed of individuals with different backgrounds and experiences, and it is difficult to separate the contribution of individuals. So while we might focus on the hitting scores of star baseball players, it is worth considering how well nine players would perform in the absence of (generally older) coaches, trainers, surgeons, and owners. In an economy, it is generally the bundle or teams that are productive, not the individuals.

For this reason, the committee tends to prefer market-based measures

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement