How does digital technology administrative penalty affect big data technology innovation: evidence from China

The rapid growth of the information technology industry has produced unprecedented volumes of data. Big data, characterized by volume, velocity, and value, underpins the digital economy and can boost productivity, operational efficiency, and long-term growth, while integrating digital and real economies. Adoption in sectors like fintech has transformed traditional models and expanded credit access. However, widespread use of big data also brings risks, including data misuse, privacy breaches, and cyber threats, which can create market failures that hinder digital innovation. Governance of these externalities is therefore essential to cultivate a supportive digital ecosystem for high-quality development. Countries seek to accelerate digital economy development through appropriate policies that manage and promote big data technology innovation (BDTI). Beyond incentives such as industrial and tax policies, regulatory systems are key to enabling innovation. Regulation can have bidirectional effects: sound digital regulation can stimulate innovation by correcting externalities and improving the environment, while inappropriate policies can raise innovation costs or create barriers. China’s rapid digital economy expansion and dynamic technological innovation provide an important context to examine whether administrative penalties—along with the public disclosure of penalty information—serve as effective tools that influence BDTI. This study fills a gap by empirically testing how digital technology administrative penalties (DTAP) imposed on firms for digital-related violations affect city-level BDTI, and by comparing regional differences to guide policy. We analyze the normative, incentive, and deterrent effects of DTAP. DTAP and penalty disclosure reduce information asymmetry and enhance the business environment; penalized firms increase long-term innovation investment to rebuild reputation; and deterrence and disclosure shape market behavior standards and competitive order. Using data from 281 Chinese cities (2008–2020), we find DTAP is positively correlated with city-level BDTI. This result is robust to alternative measures, endogeneity adjustments, control variables, time subsamples, and model specifications. DTAP also promotes new digital business forms and industrial digitalization, which in turn foster BDTI. Effects are stronger in Beijing, Shanghai, Guangzhou, and Shenzhen, and in cities with relatively low levels of digital factor-driven industry development. The study contributes to literature on digital innovation drivers, highlighting an institutional enforcement perspective and extending evidence on administrative penalties to include non-listed firms.

The paper situates its contribution within two major strands of digital innovation research: (1) firm-internal drivers (e.g., human capital, executive digital expertise, inter-firm communication, innovation alliances, and M&A in digital industries) and (2) external environmental drivers (e.g., market demand, digital infrastructure, industrial structure, foreign direct investment). Government roles previously examined include industrial policy, infrastructure, financial incentives, pilot programs (e.g., Big Data pilot zones), intellectual property rights (IPR) protection, and environmental regulation. Few studies adopt an institutional perspective focusing on administrative penalties. This study adds evidence on how administrative penalties—core regulatory tools—affect innovation, extending beyond listed firms by using administrative penalty announcements from multiple departments to include non-listed enterprises. It also contributes to enforcement mechanism literature by examining penalties as institutional enforcement that shapes incentives, reputations, and market discipline, thereby affecting BDTI.

Empirical strategy: The study employs a panel of 281 Chinese cities from 2008 to 2020 (3,350 city-year observations) and estimates two-way fixed effects models with city and year fixed effects to control for time-invariant city characteristics and nationwide annual shocks. The baseline specification relates city-level BDTI to DTAP intensity, with a vector of controls. To investigate mechanisms, mediating variable models are estimated for new business development (NBUS) and industrial digitization (IDIG), following a standard three-equation mediation approach. Identification and endogeneity: Potential bidirectional causality (e.g., higher BDTI leading to more penalties due to compliance scrutiny) and measurement error from keyword-based penalty identification motivate instrumenting DTAP with the contemporaneous average DTAP of other cities (ivDTAP) following Laeven and Levine. For mediation, NBUS and IDIG are also instrumented with the contemporaneous averages in other cities (ivNBUS, ivIDIG). Instrument validity tests are reported as passed. Variables: Dependent variables: BDTI is the number of Big Data patent applications per capita by city; rBDTI uses granted patents per capita. Big Data patents are identified on Patsnap (titles/abstracts include “Big Data”) from China’s National IP Administration. Key independent variables: DTAP is the number of administrative penalties per capita for firms associated with digital technologies (AI, AR, Big Data, Blockchain, Cloud, IoT, Metaverse, Quantum Computing, VR, 5G), identified from Shanghai Da Zhi Hui Cai Hui Data Technology Co., Ltd by firm name keywords. rDTAP is the total penalty amount per capita. Mediators: NBUS is the number of enterprises whose business scope includes keywords associated with new digital economy formats (e.g., digital governance, digital transformation, data circulation, industrial platform, Internet healthcare, micro economy, multi-point practice, online education, online office, sharing, virtual industry) per capita. IDIG is an industrial digitization index derived via factor analysis from seven indicators: counts of universities and majors for traditional and emerging digital talent, and counts of digital enterprises in primary, secondary, and tertiary sectors. Factor analysis diagnostics: KMO = 0.682; Bartlett’s chi-squared = 42532.726, p < 0.0001. IDIG is normalized to a 0–10 scale. Controls: Economic development (log real per capita GDP), GDP growth, FDI/GDP, industrial structure index, fiscal science and technology expenditure/GDP, urbanization rate, population density (log), financial depth (loan balance/GDP), and financial efficiency (loans/deposits). Robustness includes adding the intensity of “entrepreneurship” policies (POLY, log(count+1)). Data sources: China City Statistical Yearbook (city-level socioeconomics; foreign direct investment in 2020 linearly interpolated), Patsnap (patents), Shanghai Da Zhi Hui Cai Hui Data Technology (administrative penalties; digital industry indices), Bailu Thinktank (entrepreneurship policy counts). Preprocessing includes natural logs where applicable and winsorizing continuous variables at the 1% tails. Summary statistics indicate substantial cross-city variation (e.g., BDTI mean 0.5398, max 14.3056; DTAP mean 0.5110, max 16.9715). Additional analyses: Robustness checks include instrumental variables (2SLS/MLE), alternative dependent/independent measures (rBDTI, rDTAP), inclusion of POLY, exclusion of pre-2012 observations (post-2012 digital economy national strategy), and alternative estimators (random effects, pooled OLS). Heterogeneity is assessed by first-tier cities (Beijing, Shanghai, Guangzhou, Shenzhen) vs. others and by relative development of the digital factor-driven industry (constructed via factor analysis and split at annual medians).

- Descriptive and correlations: DTAP is positively correlated with BDTI (corr = 0.4040, p < 0.0001). DTAP correlates with NBUS (0.4860) and IDIG (0.2553), and both mediators correlate positively with BDTI (NBUS: 0.3896; IDIG: 0.6475), all p < 0.0001. - Univariate relationships: Scatterplots show clear linear relationships; R-squared values range from about 10.2% to 16.3% in simple regressions of BDTI or rBDTI on DTAP or rDTAP. - Baseline FE estimates: DTAP has a positive and statistically significant association with BDTI and rBDTI across specifications with and without controls. Representative coefficients from Table 4 and robustness tables indicate positive effects. For example, with instrumental variables (addressing endogeneity), DTAP → BDTI ≈ 0.156 (p < 0.01) and DTAP → rBDTI ≈ 0.031 (p < 0.01). - Robustness: Results remain after instrumenting DTAP (and using rDTAP), adding entrepreneurship policy controls (POLY), restricting the sample to post-2012, and using alternative estimators (RE, pooled OLS). Instrument validity tests are reported as passed. - Mechanism: New business development (NBUS) mediates the DTAP–BDTI link. IV mediation results: DTAP → NBUS ≈ 0.150 (p < 0.05); NBUS → BDTI ≈ 0.241 (p < 0.01); DTAP retains a positive direct effect on BDTI ≈ 0.120 (p < 0.05). Industrial digitization (IDIG) also mediates: DTAP → IDIG ≈ 0.046 (p < 0.05); IDIG → BDTI ≈ 1.084 (p < 0.01), with a remaining positive direct DTAP → BDTI ≈ 0.106 (p < 0.05). Similar patterns hold using rDTAP and rBDTI. - Heterogeneity by city tier: Effects are stronger in first-tier cities. For BDTI, DTAP coefficients are larger in first-tier cities (≈ 0.420) than in non-first-tier cities (≈ 0.148), both p < 0.01. Using rDTAP, first-tier effects (≈ 0.257) exceed non-first-tier (≈ 0.042), both significant. - Heterogeneity by digital factor-driven development: DTAP promotes BDTI in both high and low development groups, with greater impact where the digital factor-driven industry is less developed. For example, DTAP → BDTI is ≈ 0.217 (low) vs. ≈ 0.179 (high), both p < 0.01. Overall, DTAP’s normative, incentive, and deterrence effects improve the business environment, spur long-term innovation investment, and foster competitive order, thereby facilitating BDTI directly and via growth of new digital business models and industrial digitalization.

The results confirm the core hypothesis that administrative penalties can facilitate big data technology innovation by shaping a transparent and predictable institutional environment. DTAP’s normative effect reduces information asymmetry and articulates behavioral boundaries, the incentive effect pushes penalized firms to improve governance and invest in long-term innovation to restore reputation, and the deterrent effect disciplines market behavior. Together, these mechanisms strengthen confidence in the protection of innovation returns, reduce uncertainty, and lower coordination and transaction costs, thus supporting BDTI. The mediation findings show that DTAP stimulates the emergence of new digital business forms and accelerates industrial digitalization, both of which expand application scenarios, foster clustering and synergies, and provide data resources and demand that further drive BDTI. The stronger effects in first-tier cities align with their deeper financial, human capital, and infrastructure endowments, which amplify the benefits of improved regulatory enforcement. Conversely, in regions with less developed digital factor-driven industries, DTAP plays a relatively larger role by curbing disorderly behavior and creating space for new entrants and innovation. These findings emphasize the importance of regulatory quality and enforcement as complements to industrial policies, infrastructure investments, and IPR protection. Effective administrative enforcement provides policy signals, enhances market discipline, and supports collaboration between firms and government, thereby promoting innovation outcomes in the digital economy.

This study provides empirical evidence that digital technology administrative penalties (DTAP) promote city-level big data technology innovation (BDTI) in China. Using a panel of 281 cities over 2008–2020, two-way fixed effects models, instrumental variables, and multiple robustness checks show that higher DTAP intensity is associated with greater BDTI. Mechanism tests indicate that DTAP fosters new business models in the digital economy and accelerates industrial digitalization, both of which mediate the DTAP–BDTI relationship. The effects are stronger in first-tier cities and in regions where the digital factor-driven industry is at an early stage. Policy implications include strengthening the soft institutional environment (e.g., IPR protection, data governance standards, enforcement against unfair competition), improving disclosure and enforcement mechanisms for administrative penalties, and combining administrative tools with industrial policies and financial incentives to form a dual engine of technological and institutional innovation. Future research should explore heterogeneous impacts across specific digital technologies, cross-country institutional differences, and differential effects of penalty types to refine understanding of enforcement mechanisms in digital innovation.

- Technology heterogeneity: DTAP’s impact may differ across specific digital technologies (e.g., AI, Big Data, IoT), with distinct diffusion and application mechanisms not separately analyzed here. - Cross-country generalizability: Findings are based on Chinese cities; institutional environments, economic development levels, and BDTI stages differ across countries, potentially affecting external validity. - Penalty type heterogeneity: Administrative penalties vary by type and severity; the study does not differentiate their distinct effects on innovation. - Measurement limitations: DTAP intensity is identified via keyword-based matching, which may introduce classification error despite instrumenting to mitigate endogeneity. - Potential remaining endogeneity: Although instrumental variables and extensive controls are used, unobserved time-varying factors could still influence results.