Working paper no. 2000-11
This paper has substantive and methodological goals. The substantive goals are motivated by two questions: How are occupations graded by age in the United States? How has this changed over the last fifty years? Methodologically, I compare results from two different means of measuring age segregation: the index of dissimilarity and log-multiplicative models.
I find that occupations are age segregated. The best-fitting model suggests that there are two different patterns of occupational age segregation. One pattern is linear, in which workers at a particular age are progressively more segregated from workers who are more distant in age. The other pattern is u-shaped, in which workers in their twenties and sixties are more likely to perform the same occupations, and less likely to share occupations with workers at mid-life.
Among both men and women, occupational age segregation has declined during the last half of the twentieth century. The bulk of the change between 1950 and 1990 comes from changes in occupations. For example, some occupations were dominated by young workers and now are dominated by older ones. The distances separating different age groups from each other have also shifted.
The choice of method affects the results. The index of dissimilarity conflates the changes in both dimensions governing the association between age and occupation. Log-multiplicative models allow a closer examination of the particular patterns of segregation, as well as the contributions of particular occupations and age groups to those patterns.