The non‐linear impacts of innovation on unemployment: Evidence from panel data

This article examines the linear and non-linear impacts of innovation on unemployment in 61 countries, covering annual data from 2007 to 2016. The relationship is estimated by deploying the system generalised method of moments (SGMM) estimation. Alternative instruments for the SGMM estimation and cross-sectional threshold estimation are utilised to investigate the estimation robustness. Although the negative linear relationship between innovation and unemployment is not robustly supported in the linear models, the empirical results of non-linear models suggest the existence of an inverted-U effect of innovation on unemployment rate. The marginal impact of innovation also supports the inverted-U relationship in which only the innovation at the maximum and mean level contributes to lower unemployment. Additionally, the negative marginal effect is larger at the maximum level compared to the mean level. The robustness estimations that use alternative instruments and the Hansen threshold regression also support the non-linear relationship. On the basis of the findings, although innovation should be encouraged to create more jobs, policymakers are advised to consider interventions in job protection, such as offering reskilling programs to mitigate rising unemployment following the initial efforts to promote innovation.