Automation and careers
A frequent concern in media relates to the growing automation of production and the threat of a future where human work will be redundant. Such conclusions were supported by part of the scientific literature, such as Frey and Osborne (2013), though most of the field agrees that new technologies also help to create new work opportunities, e.g. community managers, content reviewer, etc. Less clear is whether workers whose jobs were subject to automation will be fit for jobs were demand is booming. Theoretical models on this issue are scarce, as empirical evidence. The research presented at the Warsaw Economic Seminars addresses the second concern
The scarce insights from theory suggest that workers leaving routine occupations faced a complex trade-off when deciding where to look for employment. In Jaimovich and Siu (2012) workers switch towards non-routine occupations due to higher expected wages, though such switch involves a longer unemployment spell that reflects the need to acquire new skills. Carrillo-Tudela and Visschers (2013) also add that workers switching occupations lose part of their accumulated human capital, which makes them more prone to lose their employments even after a successful transition. To the best of our knowledge, ours is the first analysis to test these models with individual level data.
The results of our analysis suggests that the mechanisms suggested by the models appear to exist in the data. However, the links are rather weak and country specific. Workers in routine jobs experienced more career instability in Great Britain (consistent with Carrillo-Tudela and Visschers (2013)); whereas in Germany, we observed longer non-employment spells for workers leaving jobs prone to automation. The results might reflect the characteristics of the institutional setup of both countries, particularly unemployment benefits and educational policies.
The finding that, in spite of theoretical considerations, workers from jobs prone to automation experience only slightly worse career outcomes asks for a reconsideration of the mechanisms envisioned in the models. A point that holds promise refers to the approach used to include technological progress. Models presented above suggest that all workers in non-routine jobs experience the same increase in productivity as a result of technological progress, which appears to be an unrealistic assumption. Instead, one could consider models of embedded technological progress, where only most recent jobs benefit from cutting edge technologies and thus greater productivity. Such a modification could bring job destruction in routine and non-routine jobs closer together and explain why differences are not as large as expected.