Unraveling the impact of digital transformation on green innovation through microdata and machine learning

Abstract

How to use digitalization to support the green transformation of organizations has drawn much attention based on the rapid development of digitalization. However, digital transformation (DT) may be hindered by the “IT productivity paradox.” Exploring the influence of DT on green innovation, we analyze panel data encompassing A-share listed companies in Shanghai and Shenzhen spanning the period from 2010 to 2018. It tests the DT's non-linear impact, employing a random-forest and mediation effect models. The results reveal that (i) DT can promote green innovation; (ii) regarding heterogeneity, the promotion effect is mainly manifested in enterprises in non-state-owned and highly competitive industries; (iii) based on mechanism testing, DT relies on two routes to encourage green innovation: improving environmental information disclosure and reducing environmental uncertainty; and (iv) random-forest analysis shows that DT exhibits an inverted U-shaped non-linear effect on green innovation, including the “IT productivity paradox.” This study enhances the existing discourse on DT and green innovation by furnishing empirical substantiation for the non-linear influence exerted by DT on green innovation. Furthermore, it imparts insights into the mechanisms and contextual limitations governing this association.