Trier and the task content of jobs
Just as April comes to an end, we start the conference season at Trier, where the Institute for Labour Law and Industrial Relations in the EU organized the Workshop on Labour Economics . At the conference, Lucas presented a chapter of his PhD that we expect it will soon become an independent paper.
Our research provides a new view into Routine Bias Technological Change. According to the main theory, the increase in automation can help to explain changes in employment structure. Yet, there is little evidence that other implications of the theory also hold. Particularly, most models provide a simple characterization of tasks, that compares two types of tasks: routine, where workers' productivity have little influence on outcomes (e.g. because the pace of work is determined by machines); and non-routine, which are usually characterized as requiring creativity and interpersonal skills and where workers are worse substitutes for each other. A logical implication of such description is that occupations with a greater share of non-routine tasks should present greater wage dispersion. Our research tests the validity of this statement.
By an large, data confirm the hypothesis. Occupations with greater non-routine content display greater wage inequality. This relation appears both at the top (90/50) and at the bottom of the distribution. The effect size is not large though. Increasing the non-routine task content by one standard deviation is correlated with a 10% increase in the ratio between workers at the 90 and 10 percentiles.
Since we were interested in showing that the relation arises as a result of the type of tasks that these workers supply, we had to discard alternative hypotheses. In particular, we study two alternative explanations. First, the growing demand for non-routine tasks could lead to a growing number of bad matches in those tasks, that is each new worker that provides these task is less productive than the previous. Second, it is also possible that new technologies might winner-take-all markets. While there is some evidence that wage dispersion within occupation is related to employment change and winner-takes-all, the effects of tasks remain significant, though coefficients shrank.
Evidence then confirms our hypothesis. This results provides additional confidence on the description of tasks offered by the literature. The slides are below. A working paper version of the research should be made available soon.