Occupational structure and growing wage inequality in the U.S., 1983 - 2002

Since the 1980's, wage inequality in the U.S. has been dramatically increasing. I investigate the impact of occupational structure, measured at the three-digit level, on this trend of growing wage inequality. The investigation is conducted in terms of three major research tasks. First, I test the validity of the 'disaggregate structuration' view in relation to growing wage inequality. The 'disaggregate structuration' view is suggested as an alternative to big class theories. Theorists of the 'disaggregate structuration' view assert that an occupation is a gemeinschaftlich community characterized by internal homogeneity. Thus, this view implies that most of the rise in inequality occurs between occupations and that within-occupational inequality is actually decreasing, due to the progress of 'occupationalization.' My analyses, however, find that the majority of the growth in inequality has occurred within occupations. Secondly, I thus seek a more delineated explanation for the causes of rising within-occupational inequality. I investigate whether previously proposed hypotheses can account for this phenomenon. Hypotheses that I test include demographic change, deindustrialization, unions, insecure employment relations, increases in the return to skill, and changes of firm organizations. Although smaller than within-occupational inequality, between-occupational inequality has also been growing during this period. Thirdly, I therefore investigate the changes of between-occupational inequality. Since between- occupational inequality is a weighted sum of occupational mean wages, I examine whether the same hypotheses tested for within-occupational inequality can explain the changes in occupational mean wages over time. Using the Current Population Survey (CPS) from 1983 to 2002, I find that as within-occupational inequality has grown faster than between-occupational inequality, the direct association between occupational structure and wage inequality has declined over this period. While the importance of general skills (i.e., education) in determining workers' wages is growing, the importance of occupation-specific skills is declining. For regression models of hourly wages, the amount of R-squared increase by adding three-digit occupational codes (331 occupational dummies) in addition to general skills (5 dummies of education) has decreased for this period. Therefore, the strong version of 'aggregate structuration' and 'occupationalization' is not supported. I would like to note, however, that the R-squared of hourly wage increases jumped significantly when we use three-digit occupational codes instead of one-digit occupational codes even after adjusting for the degrees of freedom. Thus, the weak version of structuration is not rejected. For multivariate tests, inequality indexes and other variables by detailed occupation are extracted from each year's CPS and merged into one panel data file with occupation as a unit of analysis. Multi-level growth models are then estimated using detailed occupational categories as the unit of analysis in order to assess how the structural characteristics of occupations affect changes in mean wages and wage inequality over this time period. Contrary to the expectations of the skill-biased technological change hypothesis, changes in the distribution of education do not affect the growth of wage inequality within occupations. In contrast to the traditional view of unions as promoting wage equality, within-occupational inequality is increased by unionization. The increase of female labor market participation seems to pull down inequality in an occupation. Deindustrialization does not account for the rise of intra-occupational inequality, while insecure employment relations do. As expected by the organizational change view, inequality grows faster in high skill jobs and service jobs. Regarding between-occupational inequality, traditional explanations do better jobs in accounting for its change than for within-occupational inequality. Skill biased technological changes and unions have positive effects on occupational mean wages. Multi-level growth models provide additional evidence against disaggregate structuration. The disaggregate structuration view assumes that occupational common interests will be achieved as accomplishment of active occupational associations. Thus, the changes of occupational mean wage, which is a clearly common interest of members in an occupation, should be explained by occupation itself, not by other demographic and institutional variables. Contrary to this expectation, most of the within-occupational variation are not explained well by other demographic and institutional variables, including race, gender, and unions. In conclusion, although sociologists often view occupation as the back-bone of the stratification system, the rise in within-occupational inequality suggests that broader, more complex approaches may be needed in order to better explain the increasing disparity in wages. I suggest that more attention should be given to firm level studies in which changes inside and between firms are investigated.