Skip to main content

Technological Change: New Sources, Consequences, and Impact Mitigation

Periodic Reporting for period 2 - TechChange (Technological Change: New Sources, Consequences, and Impact Mitigation)

Reporting period: 2020-09-01 to 2022-02-28

This project aims to deepen our understanding of technological change, how it happens, how it propagates, and how one could potentially address the negative consequences that might arise for some parts of the population.

Work funded in this grant has been structured around four major parts.

1. The first part considers how technological change might affect the size of firms, and how that feeds back on the wages that different workers earn. The main premise here is that information and communication technology allow firms to operate on a larger scale. This means that more productive firms in particular can absorb a larger part of the workforce. When workers are heterogeneous, this can amplify the differences in wages that they obtain. The idea here is to quantify in a calibrated macro model the consequences of such an affect, while still taking account of other types of technological change such as skill-biased technological change.

2. A second question within this this grant is to understand how technological change manifests geographically, and how one could counter-act negative consequences of it. These parts try to understand the consequences and possible interventions on a regional level, and is grouped around two subparts.
- 2.a If one considers national changes to industry employment, one can in principle predict employment changes at the regional level: One can weigh each employment change by the fraction of the population that is employed in this industry in a region. We study whether this is a good assumption, and if not, whether we might expect more or less dispersion than is predicted by this. We also aim to study if the local distribution of economic activity is important for aggregate effects.
- 2.b If some regions are particularly hit by technological change, we aim to investigate whether one can aid such regions, and which trade-offs might be involved. In particular, we aim to study whether industry policy that generates long-run growth is detrimental to the employment of the current workforce, while industry policy that attracts firms that readily absorb the current workforce are less effective in generating long-run growth

3. A third component under this grant aims to consider whether one can forecast parts of technological change by considering differences between firms. In particular, some part of technological progress might already be embedded in some firms and not in others, and extracting the component that already exists but is not yet widely adopted might generate insights on medium-run wide-spread technological change.

4. A final component that was added to this project concerns the implications of Covid-19, and the role of technology in it. It aims to build a model where individuals can protect themselves but the government can also intervene, to study how much old and young individuals will privately protect themselves and in which way the government should intervene on each of these groups. An important component to keep in mind here is that teleworking makes it easier for people to protect themselves and to still earn a living, which will be explicitly modeled and assessed.
For the first part (part 1 above) the project team set up a computational assignment model where firms of different productivity can hire workers of different skills. Initially the firms were restricted to hire only one worker skill but choose the number of workers, subject to decreasing returns to scale. This has been extended to allow for two types of workers (e.g. blue and white collar) where the firms can choose the skill and quantity for each. This requires to solve a differential equation system where the end-points are determined by a fixed point, which is computationally challenging. We aim to calibrate this framework by matching the size distribution, profit distribution, and the relation of wages to profits in German matched employer-employee data. The goal is to quantify the role of firms that get larger separately from other technological change. We set up the theoretical and computational model and moments from the data to calibrate it, and have successfully calibrated a simplified version.

For part 2a above we related predictions to actual employment changes in the US. So far we find that the local distribution does not seem to affect the aggregates much. But that dispersion is much larger than predicted by a simple break-down. We currently assess robustness.

For part 2b we obtained access to data from France to understand how shocks to the local market propagate. We managed to find a source of plausibly-random shocks, and are in the process of implementing it empirically.

For part 3 we sort firms in French data, and use them differentially to predict future employment consequences. This part is currently in the estimation stage on a restricted sample of data. Once the estimation is consolidated, we will use the remaining years of data to understand the predictive power of the approach.

For part 4 we developed a model of Covid-19 spread where individuals take actions to protect themselves. These actions are individually optimal, but there are externalities (if an individual becomes more risky, that affects not only the individual but also all others who interact with this individual). We calibrated the model and show that individual precautions reduce deaths a lot, but lockdowns are still beneficial if they are strong and long, and especially if they are targeted towards the young.
For project 1 (see explanation above) the progress beyond the state of the art is to consider how firm effects in terms of firm size feed back on the inequality among the workers. Preliminary results indicate that the effects are large, but might go in favor of less inequality. This arises with one worker type per firm. We expect interesting and more realistic insights to arise in the model with blue and white collar workers, and to be able to assess the economic magnitudes by the end of the project.

For project 2a the main novelty is to assess more carefully the breakdown of shocks to the regional level, and to understand the agglomeration effects and potential aggregate effects that arise. If current results are robust, it seems as if regional activity is more volatile than their employment-weighted national averages, and that this could be an important amplification for regional inequality. We do not find much affect that the distribution matters for aggregate efficiency.

For project 2b we expect to learn whether at the local level there is a trade-off between long-term growth and short-term employment. This will affect industry policy at the local level.

For project 3 we intend to learn to which extent novel technology that will affect employment over the next 2 decades is already embedded in some firms today, and whether one can extract this to predict future employment changes. We will compare this method to other methods that predict future technological change, for example based on how “routine” the tasks of a job are.

For project 4 we have already generated and published the largest insights, both on how to model economic behavior in a pandemic, its consequences, and the differences in externalities between different groups of people. The main task will be to assess how far a model like this can generate the actual evolution of the pandemic, which will be assessed by the end of the project term.