Main article

Liang Xu
School of Management, Zhejiang University, Hangzhou 310058, China
Yihan Zhao
School of Management, Zhejiang University, Hangzhou 310058, China
Mingze Chen*
School of Economics and Management, Southeast University, Nanjing 211189, China
mingze.chen@seu.edu.cn

Abstract

Digital transformation has become a central route through which firms search for new knowledge, reconfigure production processes, and accelerate the conversion of research and development inputs into valuable innovation outputs. Yet empirical evidence remains mixed because digitalization may both improve information processing and intensify short-term managerial pressure. This study develops a panel empirical and machine learning framework to evaluate how digital transformation affects innovation efficiency among Chinese A-share listed firms. Using 30,842 firm-year observations from 2010 to 2023, we construct a text-mined digital transformation index from annual reports, measure innovation efficiency using a combined DEA and patent-R&D ratio approach, and estimate the relationship through firm and year fixed effects, instrumental variables, dynamic system GMM, double machine learning, and causal forests. The baseline estimates show that a one-standard-deviation increase in the digital transformation index is associated with a 0.018 increase in innovation efficiency, equivalent to approximately 13.4% of the sample mean. Mechanism tests indicate that the effect operates through higher R&D intensity, greater knowledge recombination breadth, and improved absorptive capacity. Machine learning validation further shows that gradient boosting explains 73.1% of out-of-sample variation, and SHAP interpretation identifies digital transformation as the largest non-patent predictor. Heterogeneity analysis reveals stronger effects for non-state-owned firms, high-technology manufacturers, and enterprises located in eastern provinces. Robustness checks using alternative patent-quality weights, two-year lag structures, propensity score matching, and placebo text dictionaries confirm the main inference. The findings provide empirical support for a capability-based view of digital transformation and offer policy guidance for data infrastructure, digital talent, and innovation governance.

Article details

How to Cite

Xu, L., Zhao, Y., & Chen, M. (2024). Digital Transformation and Innovation Efficiency: An Empirical Study. Journal of Business and Data Analytics, 2(1), 1-20. https://doi.org/10.63646/j.jbda.2024.020101

Similar Articles

You may also start an advanced similarity search for this article.