Digitalization Paths, Heterogeneity and Productivity Growth —A Hidden Classes Linear Mixed Model Approach

From the perspective of digital transformation trajectory, the Latent Class Linear Mixed Model (LCLMM) was first applied in the field of enterprise management. Based on panel data of 790 small and medium-sized enterprises (SMEs) in China from 2013 to 2022, the internal mechanism of how digital transformation of SMEs affects total factor productivity (TFP) was explored. The research results show: (1) There are three trajectories of digital transformation for SMEs: speed adjustment, steady growth, and rapid growth. (2) Regarding the impact of digital transformation on total factor productivity, the speed adjustment enterprises will suppress its improvement effect, the uniform growth enterprises will exhibit an inverted “U”-shaped relationship, and the rapid growth enterprises can strengthen this promoting effect. (3) The digital transformation of SMEs improves their production efficiency by reducing R&D investment, improving innovation efficiency and shifting the resource structure to non-strategic resources; among them, the differences in R&D investment among different trajectories of SMEs are significant, while there are no differences in the adjustment of the resource structure. The marginal contribution of this study is mainly reflected in three aspects: first, explaining the paradox of digital productivity from the perspective of dynamic trajectories; second, revealing the differential mechanism of innovation capability and resource allocation under different transformation trajectories; third, for the first time, applying the Latent Class Linear Mixed Model to the field of enterprise digital transformation.