How Technological Innovation Shapes the Digital Transformation-Performance Link: Engagement and Well-Being Pathways in Chinese State-Owned Enterprises
Abstract
Digital transformation may offer employees useful work resources while introducing additional demands. Drawing on conservation of resources theory, this study examines individual-level associations between perceived digital transformation and employee performance, with employee engagement and psychological well-being as parallel statistical mediators and perceived technological innovation as a moderator. Cross-sectional survey data were obtained from 500 employees in 58 state-owned enterprises in Guangxi, China. Regression analyses with HC3 robust standard errors and bootstrap confidence intervals indicated a positive bivariate association between perceived digital transformation and employee performance. Positive conditional indirect associations were observed through both employee engagement and psychological well-being. Perceived technological innovation positively moderated the conditional direct association and both first-stage associations, although the incremental interaction magnitudes were small. Johnson–Neyman analysis suggested a negative conditional direct association in the very low moderator region, containing 51 respondents (10.2%), and a positive association in the upper region, containing 321 respondents (64.2%). These sample-specific findings warrant cautious interpretation, particularly in the sparsely represented lower region. The findings are consistent with a COR-informed account of perceived digital resources, engagement, and well-being, but do not establish causal mediation, temporal resource-conversion processes, or differences in organizational capabilities. They inform discussions of human sustainability in digital work by considering employee well-being alongside performance. Managerial implications emphasize assessing employees’ experiences of digital initiatives and innovation. Single-source measurement, nonprobability sampling, and unmodelled organizational clustering limit inference.