New projects

THE DIFFUSION OF ROBOTICS IN HUNGARY: A FIRM-LEVEL VIEW

Gábor Békés (PI), joint project with Rosario Crinò (University of Bergamo and CEPR), Alessandra Bonfiglioli (University of Bergamo, Queen Mary University London, and CEPR), and Gino Gancia (University of Milan, Queen Mary University London, and CEPR).

 The diffusion of technologies such as robotics and Artificial Intelligence is rapidly changing the way in which goods and services are produced and delivered. While automotive is still leading the automation race, other industries, including services, have started to make significant strides. As a result, nowadays robots are used not just in the assembly line, but also to carry out a variety of tasks, including those performed by warehouse handlers, pickers and packers, and more recently even bar tenders. The automation of production could have far-reaching consequences both for the competitiveness of firms and industries, and for the future of work. However, due to the scarcity of data, we still have a very limited understanding of how these technologies are adopted, why their use is concentrated in a minority of large and highly productive firms, and how they may affect firms and workers. One of the main challenges in studying automation is how to measure it, especially at the firm level. Surveys are still very limited in size, time span and coverage. As an alternative, robot imports have been used in some papers, but they suffer from two related limitations. The first is that some importers are distributors or integrators rather than final users. The second is that several (possibly many) firms purchase robots from these distributors/integrators rather than importing them directly. Hence, even if all robots were imported, robot imports at firm level alone would miscount robot adoption.

The main contributions of this project will be to build a novel and accurate measure of robot adoption at the firm level, and to use it for studying the effects of automation on Hungarian firms and workers. Our first step will be to identify robot integrators using information about the products they sell and import. Since integrators are highly specialized, we will then use firm-to-firm trade data to identify firms buying from these integrators and to trace robot imports from distributors/integrators to final users. Using this information, we will address three main issues. First, we will accurately document the process of firm-level robot adoption in Hungary and we will assess, for the first time, the quantitative importance of integrators in facilitating the diffusion of automation technologies. Second, we will identify shocks to the supply of robots in foreign countries, map these shocks to linked integrators, and trace these shocks to firms buying robots from integrators. Using these shocks, we will study how automation affects various firm-level outcomes, such as productivity, employment, occupations and markups. In addition, we will leverage occupational data linked to firms for studying which jobs are displaced by robots and, on the other hand, which professions facilitate adoption.  Third, we will analyse the consequences of automation at the broader industry level. To this purpose, we will study how automation by robot adopters affects other firms that either compete with them or are linked to them as suppliers or customers.

To carry out the project, we will combine various firm-level datasets. First, we plan to use data on firm-to-firm trade to (i) track firms’ purchases from robot distributors/integrators and use them, along with information on robot imports, to accurately measure robot adoption at the firm level; (ii) document the role of robot distributors/integrators and understand how shocks to the supply of robots in foreign countries propagate in the Hungarian economy through linkages between these firms and final robot users; and (iii) study how the effects of robot adoption spill over to non-adopters that are connected with robot users along the supply chain. Second, we plan to use customs data at the product-country level to measure robot imports by each Hungarian firm and to control for other forms of firm involvement in international trade. Third, we will need product-level sales data to identify Hungarian firms selling robots; this information will be used, along with data on robot imports, to precisely pin down the set of robot distributors/integrators operating in Hungary. Fourth, balance-sheet data will be used to construct the main outcomes at the firm level, including sales, labor and total factor productivity, total employment and estimates of markups. Finally, we plan to use matched employer-employee data to study which occupational structures facilitate or hinder robot adoption, to investigate which jobs are displaced by robots and, thanks to the possibility of following individual workers over a long time period, to study how workers displaced by automation fare many years after displacement.

Hungary is an ideal case for studying the diffusion of robotics at the firm level. While still lagging behind other countries in terms of robot density, the average number of automated units per 10,000 employees jumped from 18 to 84 over the period from 2010 to 2018 (source: International Federation of Robotics). Moreover, the presence of a large manufacturing sector implies a strong scope for robot adoption. Hence, compared to other countries, robotization is happening at a fast pace in Hungary and has a high potential for transforming the economy. At the same time, Hungary is not an important manufacturer of automation technologies, so robots are almost entirely purchased from foreign suppliers making import data suitable to measure adoption. The project aims at providing a precise quantification and an accurate description of the adoption of robotics technologies at the firm level, as well as a detailed analysis of the effects of automation on firm efficiency, market power, labor demand and workers’ employment histories. By doing so, the project will inform policy makers on how to design appropriate interventions to maximize the potential benefits of automation for the Hungarian economy while at the same time mitigating the adjustment costs for the most exposed workers. Although Hungary is an interesting case study, automation is rapidly spreading out in most EU countries. Due to the lack of detailed and reliable micro-data, the automation phenomenon is vastly understudied, and its consequences largely unknown, also at the EU level. Thanks to the rich availability of micro data for Hungary, we will be able to uncover the key features of the diffusion of automation technologies across firms and to understand the main consequences automation may have on various dimensions of the economy. By doing so, the results of the project will be useful to inform economic policies not only in Hungary but also at the broader EU level.