Robots and employment: evidence from Italy

Abstract

Increased robot diffusion has raised concerns for its possible negative impact on employment. Following an empirical approach in line with those applied to the US and Germany with contrasting results, this paper provides evidence about the effect of robots on employment outcomes in Italy (second European economy for robot stock) from the early 1990s up to 2016, both at the local labour market (LLM) level and at the worker level. In order to purge from demand and other confounding shocks, the identification relies on an instrumental variables strategy based on robots’ sectoral growth in other European countries. No harmful impact on total employment emerges from the LLM analysis; the estimated effect is negative when limited to manufacturing employment, but its statistical significance is weak or absent once concurrent trends relating to trade and ICT are controlled for. Results at the worker level show that incumbent workers in manufacturing were not damaged on average, with an overall positive (though not large) employment effect, driven by longer working relationships with the original firm; conditional on remaining at the original firm, the impact is also positive on wages. On the other hand, robot diffusion turns out to have contributed to reshaping the sectoral distribution of the new labour force inflows towards less robot intensive industries.