For part 1 we set up an assignment model where firms of different productivity choose both the ability and quantity of their workers in each occupation (blue and white collar). We provide a simple micro-foundation and provide conditions for more productive firms to hire more able workers, which is important to match German employer-employee micro-data. We focus on decade 2005 to 2015 where we can abstract from market power since we show that measured mark-ups are remarkably constant on average and display little dispersion (though we theoretically show how to embed it).
We characterize the equilibrium via a system of differential equations coupled with boundary-value-constraints that represent market-clearing. We considered many computational approaches, as the boundary problems render this difficult. Through iteration on a shooting algorithm we managed to calibrate both time periods. We fit the data quite well, and use the model as a laboratory: Quantity-bias increases the spread of firm sizes, increases average wages, but decreases wage dispersion. The latter is especially pronounced for quantity-bias for white collar workers. We presented this at the final grant conference and in a working paper.
For part 2a we completed a working paper which shows that the distribution of employment matters both for aggregate employment and especially for regional inequality, with aggregation forces amplifying large predicted positive gains.
For part 2b we searched for a setting where the amount of subsidies remains constant but the industries that receive subsidies change. We settled on regulatory changes in France that provide local municipalities with more discretion on subsidies. Using an arguably exogeneous component, we show in matched employer-employee data that this increases subsidies to manufacturing and low-skilled services at the cost of research and development. This has lasting positive employment effects. These are concentrated at the lower end of the skill spectrum, while the very high-skill sector deteriorates. The effects seem quite persistent. We presented this at the final grant conference and in a working paper.
In part 2c) we investigated predictability of occupational change based on industry leaders and followers, with some success presented at the end-of-grant conference. Also, we analyzed how one might be able to bring the right level of experience into jobs that need such experience. We expanded insights from two-sided matching with search frictions to allow firms to post their skill requirements to allow workers to adjust their search to different skill postings. We managed to capture them in a tight theoretical model that explains the large positive findings in novel field-experimental data. This was presented in many seminars, a podcast and in a working paper.
Third, for project 3 we worked on the role of technology (e.g. teleworking) during a pandemic, with Covid-19 as lead example. Since the old protect themselves more, we show that any additional lockdowns should target predominantly the young. This is especially so for ageing, industrialized countries where some teleworking is feasible. Testing is powerful in reducing the need for lockdowns. We provide general insight for other pandemics that are yet to come. We presented this to practitioners, at interdisciplinary conferences, in non-refereed general-interest columns, and the paper has a 2nd-round revision at the Review of Economic Studies.
