Do New Digital Technologies Generate or Kill Jobs

Recent advances in the fields of robotics and artificial intelligence have raised the question of whether machines and algorithms will after all make human labour obsolete (Brynjolfsson/McAfee 2011). The debate has been fuelled by a recent series of ‘future of work’ studies according to which up to half of the workforce faces a high risk of automation in coming decades (Frey/Osborne 2017). But, most studies rely on the initial routine task intensity of occupations (Autor 2015) – only an indirect indicator of technological change; or focus only on robots (Acemoglu/Restrepo 2017) – and do not capture other new digital technologies; or neglect beneficial channels of technologies which generate jobs; or ignore that firms deal with new digital technologies differently. Our aim is to study the firm-level job creation and job destruction channels arising from new digital technologies by overcoming these shortcomings.

To estimate worker-group specific labour demand as a function of firm’s technology investments, we conduct a representative firm survey among about 2.000 firms in Germany. Within the survey, we ask firms about their technology investments between 2011 and 2016. We then link the survey data to employment biographies from social security records of all workers employed in the surveyed firms. We decompose aggregate firm-level employment into a part reflecting capital-labour substitutions and product demand effects. So, we are able to empirically quantify the relative size of job creation and job destruction channels.

Our preliminary estimates suggest that technology investments are not associated with negative aggregate employment effects as job creation and job destruction effects seem to balance out. In particular, certain occupational and task groups benefit from technology, depending on the degree of automation, whereas others workers are replaced. We also find that certain task combinations are responsible for lower and higher group-specific labour demand effects from technology investments.