Digital transformation, entrepreneurship, and disruptive innovation: evidence of corporate digitalization in China from 2010 to 2021

The paper investigates whether and how firms’ digital transformation promotes disruptive innovation and the moderating role of entrepreneurship in this relationship. Motivated by China’s post-2010 transition toward a digital economy amid slowing labor force growth, the study emphasizes that digital transformation is now a strategic necessity for firms. Existing research links digitalization to general innovation but provides limited empirical evidence on disruptive innovation specifically and often overlooks entrepreneurship as a mechanism. The authors propose that digital transformation facilitates disruptive innovation and that entrepreneurship—encompassing innovation, risk-taking, and cooperation—amplifies this effect, particularly in China’s evolving market environment. The study aims to integrate these elements into a unified framework and test them empirically using firm-level data from 2010–2021.

The literature review connects digital transformation and disruptive innovation, outlining two cores of digital transformation: technology adoption (e.g., IT in production, cloud, AI, big data, blockchain) and organizational change (business processes and models). Digital transformation enhances dynamic capabilities, resource integration, information disclosure, and financing conditions, which can support innovation. Disruptive innovation involves radical departures from existing technologies and business models and is associated with high uncertainty, high R&D intensity, and significant risk. Empirical proxies often use patent citation structures (forward and backward citations). Entrepreneurship is conceptualized around innovation, risk-taking, and cooperation, consistent with Schumpeterian innovation, Knightian risk-taking, and cooperative dimensions. In China’s dynamic, opportunity-rich market, entrepreneurship is characterized by high perceived innovativeness, opportunity capture, and risk tolerance. Hypotheses: H1: Digital transformation significantly contributes to disruptive innovation. H2: Entrepreneurship positively moderates the effect of digital transformation on disruptive innovation.

Research design and data: The sample comprises A-share listed firms in China from 2010 to 2021, excluding financial sector firms (per the 2012 CSRC industry classification). Financial and governance data come from CSMAR; MD&A texts come from Juchao. Data processing uses Python for text mining and Stata 15 for econometrics. The final balanced/unbalanced panel includes 22,200 firm-year observations after excluding ST firms and those with substantial missing data; main variables are winsorized at 1.5% tails. Variables: - Explained variable (disruptive innovation, InCitepatent): Natural logarithm of the cumulative number of forward citations within five years after patent disclosure, capturing the influence of a firm’s patented technologies. Forward citations are widely used to proxy innovation impact and radicalness. - Explanatory variable (digital transformation): Two measures. 1) DT_txt: Text-based index using MD&A analysis. The authors construct a dictionary of 76 Chinese terms across five subdimensions (AI, big data, cloud computing, blockchain, digital technology application) and compute the percentage of dictionary term frequency in each firm’s MD&A, multiplied by 100. 2) DT_num: Asset-based proxy: proportion of digital-economy-related items (e.g., software in intangible assets; electronic equipment, computers, communication equipment in fixed assets) to the combined net value of intangible and fixed assets. - Moderating variable (entrepreneurship, ENT): Composite index reflecting risk-taking, innovation, and cooperation, constructed via entropy weighting of three indicators: self-generated capital satisfaction rate [(operating cash inflow + beginning cash and cash equivalents)/current cash outflow], R&D intensity (R&D/operating cost), and a dummy for jointly applied patents. - Controls: Firm age (log listing years), state ownership (SOE dummy), firm size (log total assets), board size (log board members), board independence (independent directors/board size), ownership concentration (top 10 shareholders’ share), leverage (total liabilities/total assets), liquidity (current ratio), and profitability (ROA). Year and industry fixed effects are included. Models: - Baseline FE model testing H1: InCitepatent_it = β0 + β1 DT_txt_it + γ Controls_it + Year FE + Industry FE + ε_it. - Moderation model testing H2: InCitepatent_it = β0 + β1 DT_txt_it + β2 ENT_it + β3 (DT_txt_it × ENT_it) + γ Controls_it + Year FE + Industry FE + ε_it. Robustness and endogeneity checks: - Robustness: Replace DT_txt with DT_num; add additional controls (duality of CEO/Chair, male ratio on board, ROE). Results remain significant at 1%. - Endogeneity: Lag core explanatory and controls by one period. Lagged DT_txt remains positive and significant at 1%, supporting direction from digital transformation to disruptive innovation. Heterogeneity analyses: - Ownership: Split SOEs vs non-SOEs. - Industry: Split manufacturing vs non-manufacturing. - Life cycle: Classify firms into start-up, growth, maturity, turbulence, and recession stages using Dickinson (2011) cash flow patterns and run group regressions.

- Descriptive statistics (N=22,200): InCitepatent mean 0.446 (min 0, max 2.197); DT_txt mean 0.0633; DT_num mean 0.00228; ENT mean −12.38 (indicating generally low entrepreneurship levels). - Correlations: InCitepatent positively correlates with DT_txt (0.064***), DT_num (0.066***), and ENT (0.246***). - Baseline regression (Table 4): DT_txt coefficient ≈ 0.230–0.257, significant at 1%, indicating digital transformation significantly promotes disruptive innovation (supports H1). ROA positive; leverage negative; size and board size positive; age negative. - Robustness (Table 5): DT_num positively and significantly associated with disruptive innovation (coef ≈ 4.470***). Adding further controls leaves DT_txt effect significant at 1%. - Endogeneity (Table 6): Lagged DT_txt remains positive and significant (≈ 0.258–0.295***), supporting causal interpretation from digital transformation to disruptive innovation. - Moderation by entrepreneurship (Table 7): ENT positive (0.0773***). Interaction DT_txt × ENT = 0.0440**, indicating entrepreneurship strengthens the positive effect of digital transformation on disruptive innovation (supports H2). - Ownership heterogeneity (Table 8): Non-SOEs: DT_txt positive and significant (≈ 0.9396***), ENT positive (0.0664***), interaction positive (0.0648***). SOEs: DT_txt and interaction not significant; ENT remains positive. Thus, entrepreneurship’s moderating role and DT effect are pronounced in non-SOEs but not in SOEs. - Industry heterogeneity (Table 9): Nonmanufacturing: DT_txt positive and significant (≈ 1.0382***), ENT positive, interaction positive (0.0660***). Manufacturing: DT_txt and interaction not significant; ENT positive. Hence, effects are stronger in nonmanufacturing firms. - Life cycle heterogeneity (Table 10): DT_txt not significant for start-up; significant for growth (0.9023**), maturity (0.7198*), turbulence (0.9290*), and recession (2.2271***). Interaction significant in growth (0.0665**) and recession (0.1557**), indicating entrepreneurship’s moderating role is strongest in growth and decline phases.

The findings confirm that digital transformation enhances firms’ capacity to generate disruptive innovation, addressing the research question with firm-level evidence from China’s digital economy era. Mechanistically, digitalization reduces information frictions, integrates internal and external resources, lowers R&D costs and risks, and facilitates collaboration—conditions conducive to radical, high-impact innovation. Entrepreneurship strengthens these effects by increasing firms’ propensity to take risks, invest in R&D, and cooperate across boundaries, thereby translating digital capabilities into disruptive outcomes. Heterogeneity results suggest market orientation and flexibility mediate the benefits: non-SOEs and nonmanufacturing firms leverage entrepreneurship and digital tools more effectively due to fewer institutional constraints and greater agility, while SOEs and manufacturing firms show weaker moderation effects, likely due to structural rigidities and different incentive systems. Across firm lifecycles, digital transformation is most impactful when firms have sufficient resources and direction (growth/maturity) or when it is leveraged to reconfigure capabilities during decline, whereas early-stage resource constraints can limit the immediate disruptive impact.

This study integrates digital transformation, entrepreneurship, and disruptive innovation into a unified empirical framework using Chinese A-share firms (2010–2021). It shows that digital transformation significantly promotes disruptive innovation, with entrepreneurship positively moderating this relationship. The effects are particularly pronounced in non-SOEs, nonmanufacturing sectors, and during growth and decline lifecycle stages. Contributions include: (1) constructing text- and asset-based digitalization measures; (2) operationalizing entrepreneurship via risk-taking, innovation, and cooperation; and (3) revealing heterogeneous mechanisms across ownership, industry, and lifecycle. Managerially, firms should accelerate digital construction, cultivate entrepreneurship, and tailor strategies to ownership structure, industry context, and lifecycle stage. Future research could broaden samples across countries and periods, enrich entrepreneurship measurement with additional dimensions, and deepen theoretical modeling of the mechanisms linking digitalization, entrepreneurial behavior, and radical innovation.

- Temporal and contextual scope: The sample is limited to Chinese A-share listed firms from 2010–2021, potentially constraining generalizability across countries and time periods. - Measurement constraints: Entrepreneurship is measured using three indicators (risk-taking via self-generated capital satisfaction rate, innovation via R&D intensity, cooperation via joint patents) and entropy weighting; this may omit other relevant dimensions (e.g., proactiveness, autonomy, social responsibility) and introduce measurement error. - Text-based DT measure: MD&A-based DT_txt may reflect intentions or disclosure emphasis rather than realized digital capability; although asset-based DT_num is used for robustness, both proxies have limitations. - Causal inference: While lagged models mitigate reverse causality, unobserved time-varying factors may remain; instrumental variables or quasi-experimental designs could strengthen causal claims. - Theoretical development: Although empirical mechanisms are identified, providing a more rigorous micro-foundational theory linking digital capabilities, entrepreneurial behaviors, and disruptive outcomes remains a challenge.