**4. Research Design**

#### *4.1. Data Collecting*

The initial study sample includes all Chinese A-share (RMB ordinary stock) companies listed on the Shanghai and Shenzhen Stock Exchanges from 2011 to 2020. The patent data of this study are from the database of the State Intellectual Property Office of China, and the information on listed companies and the original data of related variables are from the China Stock Market and Accounting Research (CSMAR) database. The following approaches were used to screen the data: (1) Excluding listed companies that issued both

A-shares and B-shares since they had multiple financial sources, a complicated financial structure, and potentially inconsistent data quality. (2) Eliminating any companies with unreasonable financial data or losses that have lasted longer than two years, namely the ST (Special Treatment, that is to say, exercise additional control over the stock trading of the listed companies with abnormal financial or other conditions), \* ST (Early warning of delisting risk for stocks that have lost money for three consecutive years), and PT (Particular Transfer, that is to say, stop any trading, clear the price, and wait for delisting) samples of the companies. (3) The status of publicly traded financial corporations is not taken into consideration because they operate, manage, and innovate in ways that are distinct from real economy enterprises, making it difficult to calculate enterprise innovation and other key metrics. Among them, the classification of enterprises by industry refers to the Guidelines on the Classification of Listed Companies by Industry. (4) Eliminating enterprises with serious missing indicators and abnormal data. The data of this research on TMT heterogeneity are from the CSMAR China Listed Company Database.

The enterprise patent data used in this study to calculate the enterprise innovation quality, partner diversity, and other indicators are from the patent database of the State Intellectual Property Office of P.R. China (SIPO). In the database, the number of invention patents and practical patents applied by enterprises from 2011 to 2020 was searched with "applicant = enterprise name", and a total of 422,978 patents were retrieved (Figure 2). Then this research extracts the patent field information of each patent, such as title, application number, application date, IPC, applicant type, and inventor. On this basis, this research calculates the enterprise innovation quality, partner diversity, and the TMT technological participation of each enterprise by using Python and other tools. After the above processing, the remaining 2691 enterprises have 12,797 observations.

**Figure 2.** Distribution of annual patent applications of sample enterprises.

#### *4.2. Variable Design and Specification*

4.2.1. Independent Variable: TMT Experience Heterogeneity

Based on the research of Daellenbach et al. (1999) [39], Ston et al. (2005) [40], and Yang et al. (2020) [7], this study divides the experience heterogeneity of senior management teams into functional experience heterogeneity and industrial experience heterogeneity.

Functional Experience Heterogeneity (FEH). Firstly, this paper, which is enlightened by Tihanyi et al. (2000) [68] and Yang et al. (2020) [7] and is based on the situation of sample companies, divides the functional backgrounds of TMT members into six categories: (1) manufacturing, (2) research and development, (3) financial accounting, (4) marketing, (5) law, and (6) administrative management (including Party affairs, Communist

Youth Leagues, trade unions, etc.). Secondly, this paper uses Blau (1977)'s categorical index to calculate the TMT functional experience heterogeneity, and the formula is: HFE <sup>=</sup> <sup>1</sup> <sup>−</sup> <sup>∑</sup>*<sup>n</sup>* <sup>i</sup> *p*<sup>2</sup> *ijt*. Among them, *pijt* is the percentage of members with a type *i* functional background in TMT of *j* enterprise in year *t*, and *n* is the number of functional background categories. The value of TMT functional experience heterogeneity ranges from 0 to 1. The closer the value is to one, the higher the functional experience heterogeneity of the team.

Industrial experience heterogeneity (IEH). Referring to Yang et al. (2018) [69], firstly, this research paper divides the TMT members' industries and determines the number of various industries. Secondly, the categorical index of Blau is used to calculate the value of each categorical variable separately, and the calculation formula is: IEH <sup>=</sup> <sup>1</sup> <sup>−</sup> <sup>∑</sup>*<sup>n</sup>* <sup>i</sup> *p*<sup>2</sup> *ijt*. Among them, *pijt* is the percentage of members with a type *i* industrial background in TMT of *j* enterprise in year *t*, and n is the number of industry background categories. The value of TMT industrial background heterogeneity ranges from 0 to 1. The closer the value is to 1, the higher the industrial background heterogeneity of the team.

#### 4.2.2. Dependent Variable: Enterprise Innovation Quality (Eiq)

A patent is an important carrier of enterprise innovation achievements. Traditionally, scholars take the number of patent applications and the number of patent citations as the measurement indicators of innovation quality. With the continuous deepening of research on patent text information mining, scholars are more inclined to use the breadth of enterprise patent knowledge to measure enterprise innovation quality [44,70–72].

This paper draws on the research of Liu et al. (2020) [73] and Wu Liu et al. (2022) [74], which uses the breadth of patent knowledge to represent innovation quality. First, according to the IPC classification system, which consists of five levels, namely part, large class, small class, large group, and group, the Herfindahl–Hirschman index (HHI index) at the large group level is used to measure the knowledge breadth of each patent. The calculation formula is as follows: HHI = 1 − ∑ *α*<sup>p</sup> 2. Among them, α represents the proportion of each major group classification in the IPC classification number of patent documents, and p represents the patent number. A larger HHI means a larger difference in the IPC large group classification level, a wider range of technological fields, and higher patent quality.

As for the annual innovation quality of enterprises, this study uses the natural logarithm of the median of the enterprise's annual patent knowledge breadth index plus one to measure the innovation quality of the enterprise in that year. It should be noted that according to the provisions of Chinese Patent Law on invention, utility model and design patents as well as invention and utility models have strong novelty, creativity, and practicability, but design patents are of low quality and do not have an IPC classification system. Therefore, this study only considers invention and utility model patents when measuring enterprise innovation quality.

#### 4.2.3. Mediating Variable: Enterprise Partner Diversity (Epd)

According to the classification standards of patent applicants of the State Intellectual Property Office, the applicant types are divided into five categories: enterprises, scientific research institutions, colleges and universities, government organizations, and individuals. Based on Wang's research (2021) [75], the Blau index is used to calculate the Epd, and the formula is: Edp <sup>=</sup> <sup>1</sup> <sup>−</sup> <sup>∑</sup><sup>n</sup> <sup>i</sup> s2 <sup>i</sup> . Among them, Si represents the proportion of partner type i in the annual technological innovation process of the target enterprise, and n represents the number of partner types in the annual technological innovation portfolio of the enterprise. The Epd index ranges from 0 to 1. The larger the Epd value is, the higher the cooperation diversity of the enterprise.

#### 4.2.4. Moderating Variable: TMT Technological Participation (TMTTP)

TMTTP will not only affect the innovation quality of enterprises but will also change the relationship between TMT experience heterogeneity and Edp to a certain extent. Drawing on the research of Zeng (2012) [76], TMTTP is used as a moderating variable in this

study. The formula of TMTTP is: TMTTP = Ttm−i/Ttotal−i. Among them, Ttm−<sup>i</sup> represents the number of patents with senior executives among the inventors of patents applied by the enterprise in year i, and Ttotal−<sup>i</sup> represents the total number of the patent applications of the enterprise in that year. The TMTTP ranges from 0 to 1. The larger the value of TMTTP is, the stronger the TMT technological participation is.
