Towards a Shared Prosperity: Technology, Inequality, and Labor Market

In this work package, we aim to contribute to the literature by studying the effects of waves of automation technologies on employment and wages across different European countries. Although there has been a lot of work focusing on the US and a few other countries, the comparative perspective has been missing from this literature (Acemoglu – Restrepo, 2022). Our aim, in particular, is to see whether employment and inequality effects depend on an interaction between labor market institutions, other aspects of economies, and different types of technologies. The empirical work in Acemoglu-Restrepo (2022) has been confined to the US economy. We aim to expand their empirical exploration of the role of task displacement to European economies as it may help us understand the technological and institutional reasons for wage structure developments in Europe.

To study the effects of automation on different demographic groups defined by education, age, and gender, it will be necessary to obtain detailed microdata for major European countries on wages and employment status by industries and occupations since the mid-80s or early 90s. As national labor and earnings censuses are subject to different confidential data handling policies, we aim to enhance the research laboratory (Center for Excellence) at the Department of Economic Policy to SecureData Lab that meets the highest standards in handling confidential microdata.

In addition, we aim to add new empirical evidence on the role played by technological change in labor market polarization in Europe building upon the empirical strategy of Acemoglu and Restrepo (2020) that provide a task-based approach to study the effects of automation on the demand for skills.

Autor et al. (2022) estimate that the majority of contemporary employment is found in new job tasks added since 1940. Their analysis is confined to United States Census micro-data and similar analysis for European countries is missing. They use proxies for output-augmenting and task-automating innovations built from patent data and explore how they complement and substitute labor, and thus shift occupational demand for labor. While proxies for the adoption of automation technologies in other studies explain much of the displacement effects (Acemoglu – Restrepo, 2019; 2020; 2022, Lábaj – Vitáloš, 2020; 2021), heterogeneity in reinstatement effects and creation of new work remains a puzzle.

Our aim in this work package is to provide new insights and answer the following research questions: i) Can automation explain the growing wedge between labor productivity and wages in the European countries from a theoretical and a numerical/empirical perspective? ii) What is the role of industrial robots and other automation technologies in the increasing wedge between productivity and wages? iii) What is the role of labor market institutions in mitigating the effects from new technologies on labor market outcomes? iv) To mitigate the negative effects of new technologies on workers, which policies could bring the development of labor productivity and wages in line again?

Digitalization and automation bring uneven outcomes to different social and demographic groups. The recent research in automation and its impact on inequalities is dedicated mainly to exploring variations across different occupational groups; nevertheless, the outcomes of automation on various vulnerable groups need further attention as these cohorts may encounter multiple difficulties in reaching out to the benefits of digitalization. In this work package, we build upon the intersectional theoretical and analytical framework of vulnerability (Atewologun, 2018; Crenshaw, 1990) which enables us to examine how the interconnections and interdependencies between social, demographic, and political identities contribute to different modes of discrimination in the labor market.

Based on semi-structured qualitative inquiries, the work package aims to map out and explore the viewpoints of the key policy stakeholders, including social partners, on the impact of automation on vulnerable groups with a special focus on gender, educational attainment, and ethnicity.

A technological revolution in artificial intelligence and robotics may be transformative for economic growth and human potential. Whether that growth translates to higher living standards or better working conditions will depend on institutions of governance, public investments, education, law, and public and private leadership. Given our recent experience with technology, the need to redirect technological changes towards more inclusion has become urgent. Technology does not have a path of its own, and governments could redirect it toward creating tasks that augment and empower humans (Acemoglu, 2022).

This work package will be focused on translating new findings from WP1 – WP3 into policy actions and recommendations. Policy briefs will be disseminated to policymakers and stakeholders through the project webpage, social media, and other dissemination channels.

Over recent decades, automation technologies have displaced routine tasks traditionally performed by medium-skilled workers and have contributed to growing labor market polarization. The emergence of artificial intelligence may have altered this trajectory, expanding task substitution to include non-routine tasks carried out by high-skilled workers. By employing textual analysis and examining descriptions of technology in patent texts, we develop innovative measures of occupational exposure to robot and artificial intelligence technologies. These measures are subsequently used to investigate changes in labor and skill demand in the United Kingdom over the past decade.

Our analysis reveals that the middle segment of the income distribution is predominantly exposed to robot technologies, whereas exposure to artificial intelligence rises steadily across income percentiles. Additionally, we find that exposure to robots is highest among individuals without a high school diploma and decreases consistently with higher levels of education. In contrast, artificial intelligence automation minimally affects the same group, with significantly higher exposure observed among college graduates.

Moreover, our findings highlight asymmetrical impacts of automation technologies across skill groups. Robot automation primarily reduces demand for low-skilled workers, while AI technology diminishes demand for high-skilled workers, with consistently negative direct effects despite the presence of several offsetting mechanisms. Finally, a joint evaluation of robot and AI automation effects indicates that robot automation is positively linked to increased demand for skilled workers, whereas AI automation is only weakly associated with a reduction in demand for skilled workers. These findings point to structural transformations in the labor market, carrying significant implications for wage inequality and the future of work.

What are the environmental consequences of the growing adoption of automation technologies? To address this, we propose a production model for the automation era that takes into account emission externalities. We derive a threshold condition under which the deployment of industrial robots influences emissions. This model generates three testable hypotheses:

• i) the deployment of industrial robots leads to higher emissions on average,
• ii) as the efficiency of industrial robots increases, this effect weakens and could even become negative, and
• iii) in countries where electricity is primarily generated from (clean) renewable sources, the use of industrial robots has the potential to reduce emissions.

Empirically, our findings support these theoretical predictions, suggesting that the impact of automation on emissions is non-linear or moderated by other factors.