Occupational Skills and the Evolution of Wages

The US and many European countries are witnessing substantial changes in the wage structure (Autor et al. 2006; Dustmann et al., 2009). Previous research has focused on changing returns to education and experience (Katz and Murphy, 1992), changes in the workforce composition (Lemieux, 2006), or the decline in unionization (DiNardo et al., 1996) as possible explanations for the observed changes. Until recently, little attention has been paid to the potential role of the returns to skills. However, the demand for different skills has shown vivid dynamics both within and between industries (Autor et al., 2003).

A prominent idea in the more recent literature is that structural changes have depressed the relative demand for certain types of tasks performed on the job. A common finding is that returns to routine tasks have been decreasing over time, while returns to abstract tasks have been increasing. Yet, measuring and defining different types of task categories has been proven to be empirically difficult (Autor, 2013). In our study, we avoid potential measurement issues by using a skill approach as opposed to a task approach to characterize occupations. Instead of focusing on the tasks performed on the job, we examine the skills that individuals acquire during their education.

For our empirical analysis we use the Sample of Integrated Labour Market Biographies (SIAB), a sample of social security records in Germany from 1975 through 2008. The information on skills taught in an occupation come from the BIBB/IAB Employment Survey, a representative worker survey. We construct occupation-specific skill portfolios and identify changes in the skill portfolios over the observed time period. We then match the skill portfolios to the corresponding occupations in the SIAB, thereby following the procedure utilized in Gathmann and Schönberg (2010).

In the descriptive analysis, we provide evidence that both the wage level across occupations and the skill portfolios within the same occupation have changed substantially over the observed time period. Then, we investigate whether the observed are related. Following Firpo et al. (2011), we run regressions of wage changes on the base period wage for each percentile of the within-occupation wage distribution and link the estimated intercepts and slopes to the skill portfolios of each occupation. Our results suggest that overall skills that used to be valuable in low-wage occupations have decreased in value, while the opposite has happened in high-wage occupations. Finally, we use a decomposition method based on the re-centered influence function regression approach by Firpo et al. (2007; 2009) to explicitly quantify the contribution of the skill portfolios to changes in the wage structure.

Quantifying the contribution of single skills to the evolution of wages allows us to better understand why some occupations have experienced sharp decreases in wages, while others have experienced sharp increases. We thereby contribute to the recent literature on wage inequality. Finally, calculating skill prices can provide guidance for policy makers on how to design training curricula.