The empirical evidence on how industrial robots affect employment and wages is very mixed. A meta study can help to uncover the potentially true effect of industrial robots on labor market outcomes and to identify drivers of the heterogeneous empirical results. By means of a systematic literature research, we collected 53 papers that estimate the impact of robot adoption on wages containing 2143 estimations. We utilize partial correlation coefficients and employ precision-weighted WLS estimations to analyze the true effect of robots on wages and potential publication, omitted variable and misspecification biases in the empirical literature. We also address data dependence issues as the main source on industrial robots is the IFR data set. We observe some evidence for a publication bias in favor of negative results. Moreover, we find that more positive results can be found if a) IFR data are applied, b) a finer level of aggregation than country- or industry level is used, c) a more recent time span is analyzed, d) focusing only on labor markets in Japan or developing countries, e) the functional form is of long-differences (or stacked differences), and f) the sector is non-manufacturing. We also find evidence for job polarization and skill-biased technological change since wages are more positively affected in high-skilled occupations. Differences in the labor market impact of robotization between the US and Europe vanish when one controls for more research characteristics. 

Date: Wednesday, June 21^st

Time: 4.30 to 6 pm (CET)

Location: building D4, ground floor, room D4.0.144