https://orcid.org/0000-0002-2172-4062
![Sample our Built Environment journals, sign in here to start your access, Latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5926/TOC-Subject-Sample-2021-Built-Environment.jpg"})
ABSTRACT
During the last decade, a growing literature has been assessing the impact of industrial robots on productivity and employment. At the local level, the labour-displacing effect of robot adoption can be counterbalanced by a task reallocation between manufacturing industries characterized by similar/related occupational spaces. This reallocation process can be also influenced by the average degree of occupational complexity of the local labour markets. To test for the mediating role of similarity and complexity in the robot-employment-productivity relationship, we develop a two-step analysis mixing information from different data sources on 15 industries and 39 local labour market areas of the Emilia-Romagna region in Italy, for the period 2008–2017. Our analysis shows that higher exposure to robots does not affect the aggregate level of labour productivity, but, rather, is associated with a decline in (high routine) employment after three years. We further show that the negative relationship between robots and employment decreases, or even vanishes, in local labour markets with high cross-industry occupational similarity and low occupational or task complexity, and within industrial districts.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Correction Statement
This article has been corrected with minor changes. These changes do not impact the academic content of the article.
Notes
1 Available here: https://morningconsult.com/2021/06/21/automation-workforce-jobs-polling/.
3 Occupations are commonly used as a proxy for the skills and abilities of the employees beyond their formal education (Autor, Levy, and Murnane Citation2003; Bacolod, Blum, and Strange Citation2009; Andersson, Klaesson, and Larsson Citation2014). Accordingly, occupations are defined by a set of tasks, skills, education, and attitudes needed to perform the work. The occupational space of an entity, such as firm, industry or region, reflects the degree of variety/similarity of skills or tasks that characterize such an entity (Muneepeerakul et al. Citation2013).
4 For an in-depth discussion of the use of the AIDA Bureau van Dijk archive for economic analyses in the Italian context, see Grazzi, Piccardo, and Vergari (Citation2019).
5 We choose to use the yearly installations rather than the operational stock because our dependent variable, value-added per employee, is a flow measure.
6 Our variable slightly differs from Dauth et al. (Citation2021) and Dottori (Citation2021), who compute robot density at the LLMA level. Differently, we want to keep a sectoral variability to control for sector-specific fixed effects in the regressions and to match it with our similarity measure, which varies at sector and LLMA level.
7 We do not include a variable for tangible capital, as given by the yearly value of material assets, because it can partially overlap with our robot density variable. In any case, the results do not change if we add the value of tangible capital among the regressors.
8 Relatedness metrics better captures intra-industry occupational closeness because it is a hierarchical measure that can be used to deepen the occupational space within a single industry, while similarity is more suitable for inter-industry occupational matching because it refers to the occupational distance between each couple of sectors. We choose similarity for our analysis because robot density per sector is measured at 2-digit level and thus we cannot look at intra-industry heterogeneity in terms of robotization.
9 Since the data from SILER are anonymised with respect to the individual worker identifier, we cannot apply the methodology developed by Neffke and Henning (Citation2013) to compute skill relatedness.