Yves here. It seems intuitive that introducing robot into workplaces is meant not just to reduce labor expenditures, but also suppress wage rates. However, given the number of soi disant experts that try to contend that automation increases pay levels, it’s useful to have what should be obvious confirmed.
By Daisuke Adachi, Assistant Professor Aarhus University. Originally published at VoxEU
Industrial robots can have a transformative effect on employment and wages. This column studies how the rise of robot capital has influenced wage polarisation across different occupations in the US. It finds that the substitutability between robots and labour is heterogeneous across occupations, and particularly strong in production and material-moving jobs. Robotisation has led to slower wage growth for these occupations, and thus contributed to increasing wage polarisation. Potential policies to address these distributional effects include retraining programmes and tax policies to manage the pace of robot adoption without stifling innovation.
In recent years, the rapid integration of industrial robots into manufacturing processes has spurred significant debate among policymakers and economists. The discussion centres around the transformative effects of automation on employment and the distribution of wages, especially in economies with substantial industrial sectors like the US. This column examines how the rise of robot capital has influenced wage polarisation across different occupations from 1990 to 2007, shedding light on the mechanisms driving these changes.
The Growing Role of Robots in Industry
Industrial robots have revolutionised factory production over the past three decades, with the global robot market expanding at an annual rate of 12% (IFR 2021). These advancements, however, have not affected all workers uniformly. Concerns about the uneven distributional impacts of robotics have prompted policymakers to consider measures like taxing robot adoption to mitigate potential harms. Previous studies have examined the broader employment effects of robot penetration (e.g. Acemoglu and Restrepo 2020) and explored the implications of robot taxes (e.g. Humlum 2021). However, understanding the substitutability between robots and workers within specific occupations remains crucial for assessing the full impact of automation.
Research Overview
My research (Apache 2024) delves into the substitutability between robots and workers within various occupations, using a novel dataset on the cost of acquiring Japanese robots, referred to as the Japan Robot Shock (JRS). By developing a general equilibrium model of robot automation in a large open economy and constructing a model-implied optimal instrumental variable, I address the identification challenges associated with the correlation between automation shocks and the JRS.
My analysis reveals that the elasticity of substitution (EoS) between robots and labour is heterogeneous across occupations. In production and material-moving jobs, the elasticity of substitution is as high as 3, significantly higher than the elasticity between other capital goods and labour. This indicates that robots are much more substitutable for workers in these roles compared to substitution in other occupations. The findings suggest that robotisation has significantly contributed to wage polarisation in the US between 1990 and 2007.
Implications for Wage Distribution
The high elasticity of substitution in production-related occupations implies that robotisation has disproportionately affected workers in these fields, leading to slower wage growth compared to wage growth in other occupations. Specifically, robotisation has reduced the relative wages of workers in the middle deciles of the occupational wage distribution, contributing to a 6.4% increase in the 90-50 percentile wage ratio. This measure of wage inequality, popularised by Goos and Manning (2007) and Autor et al. (2008), highlights the growing divide between high-wage and middle-wage workers.
Comparative Analysis with Existing Literature
My study complements previous research by providing a detailed estimation of the within-occupation elasticity of substitution between robots and workers. Acemoglu and Restrepo (2020) find that regions with greater robot penetration experienced lower wage and employment growth. Humlum (2021) demonstrates significant heterogeneity in the real wage effects of robots across occupations. By focusing on the US labour market and using occupation-level robot cost data, my research provides a more nuanced understanding of the substitutability between robots and workers and its implications for the distribution of wages.
Policy Considerations
The findings underscore the need for targeted policy interventions to address the adverse distributional effects of automation. Potential measures could include retraining programmes for workers in highly vulnerable occupations, as well as tax policies that are designed to manage the pace of robot adoption without stifling innovation. Given the significant impact of robotisation on wage polarisation, policymakers must carefully consider how to balance the benefits of technological advancements with the need to ensure equitable labour market outcomes.
Conclusion
The integration of robots into industrial processes has had profound effects on the wage distribution in the US, particularly by exacerbating wage polarisation. My research highlights the importance of understanding the specific dynamics of robot-worker substitutability within occupations to fully grasp the broader economic implications of automation. As the debate on robot taxation and other policy measures continues, our findings provide valuable insights for crafting strategies that mitigate the negative impacts of automation on workers while fostering economic growth.
Authors’ note: The main research on which this column is based (Adachi 2024) first appeared as a Discussion Paper of the Research Institute of Economy, Trade and Industry (RIETI) of Japan.
See original post for references
7 comments
Comments are closed.
The fact of the matter is that as populations in many major countries around the world continue to drop through declining birth rates, that robots in factories will become a necessity and so cannot be stopped. Add in an increasing part of the population being taken out of the work force due to Long Covid and you can see that countries will have no other choice.
very good point RK – AI seems to go hand-in-glove with this appraisal of robots – this was in MIT Tech Review this morning and, at least for me, i got angrier with each paragraph that i read – whose prosperity i kept asking myself –
https://www.technologyreview.com/2024/08/20/1096733/how-to-fine-tune-ai-for-prosperity
1. “Declining birth rates?” Ahem. In places like India, birth rates are declining because people are so poor that it is physically impossible for them to have more than two kids. And demographic momentum means that the population will still roughly double, but only if they have enough food: if India drains the aquifers then it’s possible that India’s fertility rate will fall below 2, but again, only out of necessity. Remember, it’s mostly not outright famine that limits population growth, but the effects of chronic malnutrition and overall grinding subsistence level poverty. Bottom line: a huge population crushed into poverty is most definitely NOT a stimulus to make expensive capital investments in robots.
2. The idea that the robots are coming for our jobs is absurd. If robots were really having an impact on labor, surely productivity would be increasing – but overall, it is decreasing. Robots are used for specific applications that require high levels of precision, or some tasks like mass producing nails, but overall we are seeing less automation, as global supplies of cheap labor are so plentiful. Robots could make our shirts – but our shirts are sewn by hand by 50 cents an hour labor, because it’s so much cheaper and more flexible than automation.
This paper covers similar ground:
Robots and Us: The Economics of the Good Life:
https://doi.org/10.1080/00346764.2018.1432884
With the added concern of this technology (now with Ai) *improving* the fields of Art, Medicine, and Law.
As robots reduce cost of production and distribution they also reduce the cumulative buying power of customers. Capitalism is self destructive.
Continuation of an ongoing debate in society of Capital vs Labor. At least with increased automation via robotics, society would benefit not only from enhanced productivity gains but as long as the “robots” are built and produced locally, we might avoid the supply chain issues facing industry today (due to geopolitical factors). How is this any different than the NAFTA agreements signed into Law by the “progressives” in the 90’s and how has that worked out for the average American manufacturing worker? I would rather have this automation capability available domestically than to be purchasing finished goods from abroad! What about robotics in the service industry ……. Much more opportunity to flourish there IMO.
Far as I can see, once upon a time robots on the factory floor required supervision and maintenance, which meant people. And robots were made in factories where people worked. Nowadays, robots make robots, robots supervise robots, robots maintain robots. Coming soon, robots manage robotic factories and robots write articles for robot-edited newspapers about how using more robots has no effect on employment and wages.