Research data set on AI application rhythm and breakthrough innovation of China's Shanghai and Shenzhen A-share high-tech listed companies from 2011 to 2024

Research on the Mechanism and Path of Business Environment Affecting the High-Quality Development of Leading Private Manufacturing Enterprises (No. 22BGL099) and the Guangdong Natural Science Foundation

Project

Process Mechanism of Green Strategy Response of Chinese Manufacturing Enterprises Under the "Dual Carbon"

Background

Theoretical Model and Empirical Research (No. 2024A1515012805). It corresponds to the academic paper A Study on the Impact of Artificial Intelligence Application Rhythm on Corporate Radical Innovation.Time Span: The panel observation interval covers 2011 to 2024. Since the calculation of the core explanatory variable (artificial intelligence application rhythm) requires 4-year retrospective data, the underlying data of artificial intelligence application degree spans from 2007 to 2024.Sample Size: The dataset contains 1,778 high-tech listed companies in China’s Shanghai and Shenzhen A-share markets, with a total of 9,654 firm-year observations.II.

Sample Selection and Data Sources1. Sample Screening RulesTaking high-tech listed companies in China’s Shanghai and Shenzhen A-share markets from 2007 to 2024 as the initial sample, this dataset is screened according to the following criteria:Excluding samples of financial listed companies;Excluding samples with abnormal operating status such as ST and *ST firms;Excluding samples with missing key data for core variables;All continuous variables are winsorized at the 1% level on both sides to mitigate the interference of extreme values on empirical results.Since the calculation of artificial intelligence application rhythm requires 4 consecutive years of artificial intelligence application degree data, the final valid observation interval available for empirical analysis is 2011–2024.2.

Data SourcesThis dataset is integrated from multiple authoritative data sources, with details as follows:Annual reports of listed companies: Sourced from Sina Finance. They are used to extract artificial intelligence keywords for measuring the degree of corporate artificial intelligence application, and to manually identify the information technology background of top management teams.Patent data: Sourced from the China Research Data Services Platform (CNRDS).

It is used to measure the level of corporate radical innovation based on IPC classification codes.Corporate finance, governance characteristics and government subsidy data: Sourced from the China Stock Market & Accounting Research (CSMAR) Database, covering multi-dimensional indicators such as corporate assets and liabilities, ownership structure, audit fees, and financing constraints.III. Variable Definition and Measurement DescriptionThis dataset contains 21 fields, including explained variables, core explanatory variables, moderating variables, control variables, and heterogeneity grouping variables.

The specific definitions of each variable are as follows:1. Explained VariableRadical Innovation (ERI): It takes the number of patents under the new 2-digit IPC classification codes entered by the enterprise in the current year as the basic indicator. If an enterprise enters an IPC classification field that it has not involved in the previous five years in the observation year, the corresponding patents are identified as radical innovation outputs.

The number of patents is added by 1 and then taken the natural logarithm to alleviate the right-skewed distribution of the data.2. Core Explanatory VariableArtificial Intelligence Application Rhythm (AIAR): It depicts the irregularity of the enterprise’s artificial intelligence application trajectory, and is operationalized as the kurtosis value of the first-order difference of artificial intelligence application degree within a rolling 4-year window.

A larger value indicates a more irregular time distribution and stronger rhythm fluctuation of the enterprise’s artificial intelligence application.Specifically, the degree of artificial intelligence application is measured by the natural logarithm of (1 + the frequency of artificial intelligence-related keywords in the annual reports of listed companies). The rhythm calculation refers to the classic kurtosis formula in the fields of internationalization rhythm and organizational change, with a rolling window of [t−4, t].3.

Moderating VariablesManagement Information Technology Background (MITB): The proportion of executives with information technology-related education or work experience in the total number of the top management team. Educational background includes study experience in computer science, information science, big data and other related majors; work experience includes positions in information technology R&D, system development, internet-related fields, etc.Government Subsidies (GS): The ratio of the total amount of government subsidies obtained by the enterprise in the current year to its operating revenue, which is used to measure the intensity of external financial support received by the enterprise.

The government subsidy data is screened and summed from the "Details of Government Subsidies" under "Non-operating Income" in annual reports by keyword retrieval method.4. Control VariablesReferring to authoritative literature in the innovation field, the following characteristic variables affecting enterprise innovation are included:Asset-liability Ratio (LEV): Total liabilities / Total assetsListing Age (AGE): ln(Current year - Listing year + 1)Board Size (BS): Natural logarithm of the number of board directorsOwnership Concentration (TOP1): Shareholding ratio of the largest shareholderAudit Fee (AF): Natural logarithm of audit feesFinancing Constraint (FC): FC index to measure the degree of financing constraints faced by enterprisesProportion of Independent Directors (PID): Proportion of independent directors in the total number of board directors5.

Heterogeneity and Auxiliary VariablesThe dataset also includes auxiliary fields for heterogeneity analysis and robustness tests:Property right nature (State-owned enterprise dummy): Distinguishes between state-owned and non-state-owned enterprise samplesIndustry attribute (Manufacturing dummy): Distinguishes between manufacturing and non-manufacturing high-tech enterprise samplesFirm age: Used to classify mature enterprises and start-up enterprisesManagerial myopia: Measures the short-term behavior orientation of enterprise managementProportion of digital technology intangible assets: Digital technology intangible assets / Total intangible assets, used to replace the measurement indicator of artificial intelligence application degree in robustness testsProvince code and industry code: Used to control province and industry fixed effectsIV.

Data Processing InstructionsCore indicator calculation: The artificial intelligence application rhythm is calculated by the rolling window kurtosis method. Based on 5 years (t−4 to t) of artificial intelligence application degree data, the annual change is obtained through first-order difference, then the adjusted standardized kurtosis is calculated to derive the annual rhythm indicator.Panel structure: The dataset is unbalanced panel data, allowing entry and exit of enterprises due to listing, delisting or data missing.Standardization processing: All continuous variables have been bilaterally winsorized at the 1% level to ensure the robustness of empirical estimation; ratio variables involving scale have been de-scaled to enhance comparability among enterprises.V. Value and Application Scenarios of the DatasetAcademic value: This dataset introduces the "rhythm perspective" into the research scenario of artificial intelligence application for the first time, provides measurement data of artificial intelligence application from a dynamic temporal dimension, and makes up for the deficiency that existing studies mostly measure the level of artificial intelligence application from a static perspective.

It can provide data support for empirical research in fields such as artificial intelligence and enterprise innovation, strategic rhythm, and organizational behavior.Application scenarios: It can be used to test research topics such as the innovation effect of artificial intelligence application, the economic consequences of technology application rhythm, the moderating effect of executive background characteristics and policy tools, and the differences in technology empowerment under enterprise heterogeneity.

It can also serve as supplementary data for research in fields such as digital transformation and radical innovation.VI. File DescriptionThis dataset is stored in Excel format, containing 1 worksheet (Sheet1). Each row corresponds to one firm-year observation, and each column corresponds to one variable field.

The data has been cleaned, winsorized and standardized, and can be directly used for empirical research such as panel regression, heterogeneity analysis and robustness tests.