Registration System Reform and Enterprise Innovation: Evidence from a Quasi-Natural Experiment of the Registration-Based IPO System Reform Pilot in China

Abstract

In recent years, the registration-based IPO system has been gradually introduced and promoted in China’s capital market, and its implementation effect has attracted considerable attention. This paper focuses on companies that were first listed between 2019 and 2021. We empirically investigate the impact of the registration system reform on enterprise innovation using the staggered difference-in-differences model, with R&D investment as the measure of enterprise innovation. The findings demonstrate that, in comparison to companies listed via the approval system, those listed via the registration system exhibit a higher level of R&D investment in the three years following listing. This suggests that the reform of the registration system effectively stimulates an increase in R&D investment among IPO companies. Furthermore, the reform of the registration system has been found to significantly promote the R&D investment of IPO companies on the STAR Market (i.e., the Science and Technology Innovation Board), while having no significant impact on the R&D investment of IPO companies on the ChiNext Market (i.e., the Growth Enterprise Board). Further analysis indicates that the registration system reform encourages IPO firms to increase R&D investment by reducing agency costs, alleviating financing constraints, and accumulating human capital. This study elucidates the impact of registration system reform on enterprise innovation and its mechanism and provides novel empirical evidence for the evaluation of the effect of registration system reform pilot.

1. Introduction

On 5 November 2018, President Xi Jinping announced the establishment of the STAR Market and the pilot of the registration-based IPO system, thereby marking the official launch of the reform of China’s registration-based IPO regime. In the following two years, the STAR Market and the ChiNext Market were both reformed in their registration system. Following a four-year pilot phase, the full implementation of the registration-based IPO system was formally initiated on 1 February 2023. The initial public offering (IPO) system in China’s capital market has undergone two distinct phases: the approval-based IPO system (examination system; 1990–2000) and the approval-based IPO system (verification system; 2001–2019). The previous approval-based IPO system (examination system) entailed a considerable degree of pronounced administrative intervention by the government. The subsequent implementation of the approval-based IPO system (verification system) facilitated the gradual transition of initial public offerings towards a more market-oriented approach [1]. The registration-based IPO system has further enhanced the IPO competition mechanism and reinforced the regulatory function of the market. The registration-based IPO reform represents a pivotal institutional innovation in the evolution of China’s capital market development. Currently, the economic implications of this reform have emerged as a central concern, attracting the attention of a diverse range of stakeholders in the Chinese capital market.

As China’s capital market has developed and matured, initial public offerings (IPOs) have become the predominant means through which modern enterprises secure the capital they require for growth from investors in the stock market. The very nature of enterprise innovation is characterized by significant uncertainty and the inherent possibility of failure. This necessitates substantial investments of both time and capital. In this context, an IPO in the capital market can facilitate enterprise innovation [2]. Previous studies have examined the impact of going public on enterprise innovation in the context of approval-based IPO systems in China. These studies have found that IPO can promote enterprise innovation by alleviating financing constraints, standardizing internal governance mechanisms, and strengthening research and development team building [2]. In comparison to the approval-based IPO system, the registration-based IPO system is distinguished by the following characteristics: (1) The competent authorities only conduct a formal examination of the information and materials provided by the issuance applicant, focusing on the evaluation of the information disclosure quality of the application materials; (2) The range of listing options available to enterprises has become more diverse, and the establishment of listing criteria has become more aligned with the characteristics and positioning of each market segment, demonstrating a greater degree of inclusiveness; (3) A more market-oriented underwriting mechanism is implemented, whereby the pricing power is directly entrusted to the market; (4) The regulatory authorities reinforce their supervision of listed enterprises by imposing stringent requirements on intermediary institutions in fulfilling their respective responsibilities; (5) The registration-based IPO system eliminates the value judgments of administrative agencies on listed companies, and investors must rely on publicly disclosed information to make investment decisions and assume investment risks; (6) Relevant departments enforce more rigorous oversight over listed companies throughout the process of going public and post-listing. In light of the above, it can be stated that the reform of the registration system represents a comprehensive and multi-level reform of the IPO process within the context of the capital market.

The question thus arises as to whether listing through the registration-based IPO system is more conducive to promoting enterprise innovation than listing through the approval-based IPO system. The discussion on this issue has significant theoretical and practical implications. On the one hand, innovation represents a pivotal driver of national development and economic growth. As the nucleus and principal entity of the national innovation system, the capacity for innovation exhibited by enterprises has garnered considerable attention. It has gradually become a pivotal criterion for evaluating the strength of enterprises in the market. Increased investment in innovation within enterprises secures new competitive advantages, boosts corporate performance, fosters customer loyalty to products, and plays a crucial role in swiftly establishing market presence, thus substantially elevating enterprise value and facilitating sustainable development [3,4]. Exploring the impact of registration system reform on enterprise innovation will help enrich the research on the relationship between going public and innovation, reveal the differences in the effects of listing through the registration-based IPO system and the approval-based IPO system on enterprise innovation, further deepen the relevant research on the influencing factors of corporate innovation, and provide a reference for the innovation development of listed enterprises. On the other hand, as a recently implemented policy, the economic effects of the reform of the registration system have attracted much attention. Previous studies have suggested that the reform of the registration system has made important contributions to the sustainable development of the capital market, including improving the efficiency of market pricing, helping investors form expectations about the future healthy development of the market, and curbing speculative behavior [5]. At present, however, relevant research is still in its infancy, and there is a lack of literature on the impact of the registration system reform on enterprise innovation. Therefore, the in-depth study of this issue will not only enrich the research on the economic impact of the registration system reform but also provide empirical evidence to evaluate the policy effect of the pilot of the registration system reform and provide a reference for the further refinement of the registration system for share issuance. In addition, the research results can also be extended to other countries’ capital market reforms and provide references for countries that have not yet adopted a registration-based listing system.

In conclusion, based on the quasi-natural experiment of the registration system reform pilot in the STAR market and the ChiNext market, this paper takes the companies listed under the registration-based IPO system from 2019 to 2021 as the treatment group, while the control group consists of the companies listed under the approval-based IPO system in the same period, and adopts the staggered difference-in-differences (DID) model to empirically test the impact of the registration system reform on enterprise innovation. At the same time, considering the long cycle of innovation activities and the relatively short implementation period of the registration system, this paper adopts the relative scale of R&D investment as a proxy variable for enterprise innovation, referring to the similar method of Zhang [6]. The results show that, compared with listing under the approval-based IPO system, listing under the registration-based IPO system can significantly promote the increase in R&D investment, but this promotion only affects companies listed on the STAR market. In addition, the promotion effect of listing through the registration-based IPO system on enterprise innovation shows a lag effect, that is, the promotion effect is not significant in the year of listing but is significant and increases year by year in the three years after listing. Further mechanism analysis indicates that the reform of the registration system plays a more important role in promoting enterprise innovation for non-state-owned enterprises (non-SOEs) with a higher degree of financing constraints than SOEs and for enterprises with a lower proportion of technical personnel; moreover, the registration system reform helps alleviate the agency conflict between shareholders and managers and reduce the agency costs of enterprises to promote enterprise innovation. The above results demonstrate that the reduction in agency costs, the easing of financing constraints, and the accumulation of human capital are important channels through which the reform of the registration system facilitates enterprise innovation.

The research contributions of this paper include the following four aspects. First, this paper examines the impact of going public on innovation in the context of reforming the registration system and enriches and extends the literature on the relationship between going public and innovation. Existing studies mainly focus on the impact of going public on enterprise innovation from the perspective of traditional functions of going public, such as post-listing restrictions, and governance of listed firms. Few studies pay attention to the changes in the impact of listing on enterprise innovation resulting from primary market institutional changes, while this study finds that the reform of the registration system can significantly enhance the positive effect of listing on enterprise innovation.

Second, from the perspective of enterprise innovation, this paper examines the economic consequences of the registration system reform and provides new empirical evidence to evaluate the effects of the registration system reform. Our research sample covers the period from 2016 to 2022 (the data were collected from three years before and after the IPO), during which the registration system reform was gradually implemented in China’s capital market, leading to the special situation of parallel registration-based IPO system and approval-based IPO system, which provides a unique situation for studying the effect of the registration system reform pilot, so the research conclusion is of great significance at the level of institutional reform. Current research mainly examines the economic effects of the registration system reform from the perspectives of pricing efficiency, market supervision, and information disclosure, with limited research on the effects of the reform on enterprise innovation and other firm behaviors. From the perspective of enterprise innovation, this paper finds that the registration system reform can promote an increase in enterprise investment in research and development.

Third, this study further explores the mechanisms through which the registration system reform affects enterprise innovation and reveals the reasons for the difference in the impact of the registration-based IPO system and approval-based IPO system on enterprise innovation, which contributes to stakeholders’ better understanding of how the capital market system affects enterprise innovation activities. The study finds that compared with listing through the approval-based IPO system, listing through the registration-based IPO system can further reduce a firm’s agency costs, alleviate a firm’s financing constraints, and accumulate human capital to promote enterprise innovation.

Finally, given the distinctive characteristics of different market segments under the registration system, this paper discusses the differences in the impact of registration system reform on enterprise innovation between the STAR market and the ChiNext market and further provides reference suggestions for market investment and policy making.

2. Literature Review and Hypothesis Development

2.1. Literature Review

2.1.1. The Relationship between IPO and Enterprise Innovation

The impact of initial public offerings (IPOs) on enterprise innovation has been a subject of debate in the existing literature, which presents two opposing views.

One perspective suggests that IPOs hurt enterprise innovation. Ferreira et al. [7] and Gao et al. [8] started from the constraints of post-listing enterprises and found that after going public, firms face constraints such as information disclosure requirements, stringent regulations, and pressure to meet shareholders’ return expectations. These constraints may lead IPO firms to prioritize more traditional and conservative projects, resulting in a decline in innovation quality. Zuo and Zhang [9] examine companies listed on the STAR Market from 2019 to 2021 and find that companies listed on the STAR market engage in R&D “whitewashing” behavior, which is not conducive to enterprise innovation. Bernstein [10] used an instrumental variable approach to demonstrate that the transition to the public stock market led enterprises to redirect their R&D investment to more traditional projects, which led to a decline in innovation quality. At the same time, by comparing the innovation activities of listed companies with those that withdrew from the IPO and maintained their privacy, he found that the IPO led to a loss of skilled inventors and a decline in the productivity of the remaining inventors, which ultimately led to a decline in internal innovation quality. Most scholars who support this view discuss the impact of going public on enterprise innovation strategy management from the perspective of post-listing restrictions.

Another perspective suggests that IPOs can promote enterprise innovation. Most scholars supporting this view examine the impact of IPO on enterprise innovation from the perspective of traditional functions of listing and governance of listed firms. Zhang et al. [2], who analyzed manufacturing enterprises listed from 1999 to 2006, used propensity score matching (PSM) and difference-in-differences (DID) methods and found that IPO can alleviate financing constraints, strengthen corporate governance and talent team building, and further promote the significant increase in the number of enterprise innovations. Ge et al. [11], based on the high-tech enterprises in China’s manufacturing industry from 1998 to 2013, applied the PSM-DID method and found that going public can alleviate financing constraints and enhance corporate reputation, thus promoting the continuous innovation of high-tech enterprises. Bernstein [10], comparing the innovation activities of listed companies with those that withdrew their IPO applications and maintained their privacy, found that listed companies attract new human capital and obtain external patented technologies to promote external innovation. Using a multi-time DID model, Zhang et al. [6] studied companies listed on the ChiNext market from 2010 to 2016 as the treatment group and the unlisted companies that attempted to go public during this period but failed as the control group. The study suggests that IPOs can alleviate financing constraints, mitigate agency problems between managers and shareholders, and attract high-level talent, thereby encouraging firms to increase R&D investment and enhance enterprise innovation.

In conclusion, the extant literature on the relationship between listing and enterprise innovation is largely confined to an examination of the impact of listing on innovation from the perspectives of traditional listing functions, post-listing restrictions, and the governance of listed enterprises. There is a paucity of research that considers how changes in the primary market system can influence the relationship between listing and enterprise innovation.

2.1.2. The Economic Consequences of Registration System Reform

As a significant reform in China’s IPO market, the economic consequences of the registration system reform have attracted considerable attention, although research in this area is still in its infancy. Furthermore, the existing academic literature predominantly examines the effects of registration system reform from the perspectives of pricing efficiency, market supervision intensity, and information disclosure. However, there is a paucity of studies exploring the impact of registration system reform on enterprise behaviors such as innovation and research and development (R&D).

At present, there is a substantial body of research examining the impact of registration system reform on market pricing efficiency. However, the conclusions researched by different scholars on this topic vary. Lai et al. [5] investigated the market reaction and short-term performance of IPOs and found that the implementation of the registration system markedly enhanced market pricing efficiency. Liang [12] and Zhang and Teng [13] reached the conclusion, based on empirical tests, that the implementation of the registration system effectively enhances pricing efficiency and curbs IPO underpricing. However, Wu and Zhang [1], Feng et al. [14] and Li and Li [15] conducted empirical research and discovered that the registration system increased the rate of IPO underpricing.

The reform of market regulation through the registration system is mainly reflected in two main aspects: firstly, the strengthening of the inspection of the performance of intermediary agencies and, secondly, the enhancement of the post-supervision. In recent times, scholars have also undertaken research in this area, although there is a paucity of empirical studies on the subject. Luo et al. [16] posited that the registration system emphasized and consolidated the “gatekeeping” responsibility of sponsor institutions and sponsors in the initial public offering examination. Furthermore, regulatory authorities augmented the supervision of enterprises through the supervision of the performance of intermediary institutions. Concurrently, following the implementation of the registration system reform, regulatory authorities have augmented the penalties imposed on listed companies, particularly those about violations of information disclosure. This has resulted in a notable reduction in the incidence of IPO violations. The robust and efficacious back-end supervision of the registration system has served to curtail IPO violations to a considerable extent and enhance the efficiency of market supervision [17].

The extant literature on the impact of reform of the registration system on enterprise behavior is limited. Several studies have examined the influence of the registration system on the earnings management behaviors of enterprises. For example, Zeng et al. [18] found that companies listed through the registration-based IPO system exhibited a reduced degree of earnings management during the issuance period. Similarly, Huang et al. [19] demonstrated that the registration system reform of the STAR Market not only influenced the earnings management behavior of companies in the IPO process but also significantly reduced the extent of earnings management of companies in the year after the IPO. However, Shi et al. [20] discovered that shell companies tended to engage in more accrual earnings management during the period of active implementation of the registration-based IPO system. Other studies have concentrated on the disclosure behavior of firms listed through the registration system. For instance, Liu and Li [21] have indicated that the information disclosure of IPO companies on the STAR Market under the registration system will encourage companies in the same industry to increase R&D investment and enhance investment efficiency. Similarly, Yu et al. [22] employed an LDA topic model to analyze the audit inquiry letters and prospectuses of IPO companies on the STAR Market and found that the audit inquiry process within the registration-based IPO system could improve the information disclosure degree in the prospectuses, which was conducive to improving the information disclosure quality of IPO companies. Furthermore, some studies have evaluated the overall performance of listed enterprises. For example, Mao et al. [23] discovered that reforms to the registration system markedly diminished the difficulties encountered by IPO companies in sustaining their pre-IPO performance levels following listing. The extant research concerning the impact of registration system reform on enterprise behavior is both comprehensive and extensive; however, the overall attention devoted to this topic is inadequate, and the current literature is relatively scarce.

The STAR Market and the ChiNext Market represent integral components of China’s capital market system. In the period between 2019 and 2020, both boards conducted piloted studies of the registration-based IPO system. Given the distinctive features of each board, the registration-based IPO policies applicable to them also exhibit differences. At present, the academic community’s focus is predominantly on the impact of the registration system reform on the overall market economy. However, there is a relative scarcity of literature specifically analyzing the policy differences between the STAR Market’s registration system and the ChiNext Market’s registration system, as well as the differences in their implementation effects.

Some existing studies have discussed the systemic differences between the registration systems of the two boards, either directly or indirectly. Lin and Zhang [24] posit that a modern capital market is an inevitable choice to support the new economy, based on their exploration of the characteristics and differences of the coordinated advancement of the STAR Market, the ChiNext Market, and the Beijing Stock Exchange. Wang et al. [25] discuss the advantages of the ChiNext Market’s registration-based IPO system and suggest that, based on the pilot registration-based IPO system of the STAR Market and advanced foreign experience, the ChiNext Market’s registration-based IPO system should establish a coherent integration between traditional and modern systems, address the limitation of the original mechanism, and compete with the STAR Market in a differentiated manner to facilitate the advancement of capital market reforms. In his analysis of the background to the introduction of the ChiNext Market’s registration-based IPO system, Zhang [26] provides a detailed interpretation of the reforms to the registration system in the ChiNext Market. He also compares the similarities and differences between the registration-based IPO system of the STAR Market and the ChiNext Market.

A minority of scholars have paid attention to the differences in the registration-based IPO system between the STAR Market and the ChiNext Market when exploring the economic consequences of the registration system reform. Furthermore, they have conducted empirical analysis in comparison. Bian et al. [27] examine the changes in the Chinese stock market before and after the implementation of the registration-based IPO system in different market segments. They also discuss the response of listed companies to the implementation of the registration-based IPO system from two perspectives: financial risk factors and corporate equity attributes. Their findings reveal significant differences in the response of different market segments to the introduction of the registration-based IPO system. Yu et al. [28] employ manual reading methods and computer text analysis to investigate the impact of risk disclosure and text tone in the prospectus of listed companies under the registration-based IPO system on the IPO discount rate. Furthermore, they examine the influence of different market segments and new inquiry price regulations on this relationship. A heterogeneous analysis indicates that the impact of risk disclosure in the prospectus on the IPO discount rates on the STAR Market and the ChiNext Market is consistent.

In conclusion, there is a paucity of literature that explores the impact of the registration system reform on corporate governance behaviors such as corporate innovation. Moreover, in the context of economic effects associated with the registration system reform, there is a predominant focus on the overall market. Additionally, there is a dearth of literature examining the differential effects of the registration systems of the two boards.

2.2. Hypothesis Development

2.2.1. Registration System Reform and Enterprise Innovation

The process of enterprise innovation is frequently lengthy and characterized by considerable uncertainty and the potential for failure. The utilization of internal funds to foster enterprise innovation not only affects the turnover efficiency of these funds but also subjects enterprises to considerable operational risks. Conversely, supporting enterprise innovation through the external capital market presents several notable advantages. The external capital market is well-positioned to fulfill a few important functions, including the allocation of scarce resources, the reasonable pricing of innovation projects, the dispersion of risks, and the motivation and restraint of managers [29]. In the context of an approval-based IPO system, previous scholars have identified a positive correlation between listing and enterprise innovation [2,6]. The registration system reform represents a comprehensive and multi-level transformation of the IPO process within the capital market. It modifies the IPO market environment and affects a range of enterprise management activities. This paper identifies three potential mechanisms through which registration system reform may affect enterprise innovation.

• Channel 1: Reducing Agency Costs

The innovation practice of enterprises is beset with significant agency problems, largely due to the absence of immediate financial returns from R&D activities. Additionally, the professional and confidential nature of these activities further complicates the situation. Zhu et al. [30] discovered that the confidential and uncertain nature of enterprise innovation intensifies the information asymmetry between shareholders and management, leading agents to excessively focus on short-term performance targets at the expense of enterprise innovation. Moreover, Richardson [31] and Hall [32] have demonstrated that the innovation and R&D activities of enterprises require sustained long-term efforts on the part of operators, while agents often receive only a modest remuneration stipulated in the contracts. Furthermore, the ultimate beneficiaries of the transformation of innovation achievements are the enterprise owner, which will inevitably lead to the on-duty consumption of agents. If the degree of information asymmetry can be reduced, the supervision of shareholders over the management can be strengthened, and the principal-agent problem of enterprises can be mitigated, the current situation in which the management ignores corporate innovation and intentionally controls the R&D investment of enterprises can be improved to a certain extent.

In accordance with agency theory, the separation of ownership and control can give rise to several potential conflicts of interest between managers and owners, leading to the first type of agency cost [33]. In the context of new share issuance under the registration system, the relevant national regulatory authorities adopt a more relaxed approach to enterprise qualifications, utilizing market-oriented inquiry, pricing, and underwriting mechanisms. This results in the transfer of valuation and pricing authority to the market [34], creating a favorable environment for market participants to interact and encouraging shareholders to enhance risk awareness and intensify their scrutiny of management governance. It is therefore anticipated that, in comparison with the approval-based IPO system, shareholders under the registration-based IPO system will exhibit a stronger motivation to motivate and supervise the management to safeguard their interests. This will serve to further reduce the first-type agency cost. Conversely, following the theory of information asymmetry, it can be observed that shareholders possess comparatively restricted access to internal enterprise information. This may result in the consequence that the impact of corporate governance may not align to maximize shareholders’ interests, thereby giving rise to certain agency issues [35]. The registration-based IPO system imposes higher requirements for the accuracy and integrity of information disclosure. Improved information disclosure quality reduces information asymmetry, enabling shareholders to have a more comprehensive and realistic understanding of the company’s situation. This alleviates agency problems. Therefore, we argue that compared with listing through an approval-based IPO system, listing through a registration-based IPO system can further alleviate the agency problem of enterprises, reduce the first type of agency cost, and promote enterprise innovation.

• Channel 2: Alleviating financing constraints

In imperfect markets, Myers and Majluf [36] proposed the pecking order theory, which posits that information asymmetry is a significant factor contributing to enterprises’ financing constraints. Nevertheless, the rigorous standards of the registration-based IPO system for the precision and reliability of information disclosure empower investors with comprehensive insight and accurate identification of managers’ behaviors [37], thereby reducing the extent of information asymmetry. This, in turn, facilitates enterprise attraction of greater external investments, alleviating financing constraints and advancing the implementation of enterprise innovation projects. In addition, compared with developed capital markets, the approval-based IPO system in China sets a relatively high threshold for enterprises to be listed, and the China Securities Regulatory Commission (CSRC) has established relatively rigorous refinancing qualifications, making it challenging for enterprises to obtain funds through equity financing. Consequently, the distinctive institutional context imposes more pronounced constraints on financing for Chinese enterprises [38]. In the context of a registration-based IPO system, regulatory authorities shift their focus from scrutinizing enterprise qualifications to evaluating the performance of information disclosure obligations. This shift empowers the market with greater choices and pricing power, thereby providing enterprises with increased opportunities to access the capital market, particularly benefiting small-scale companies [39]. This expansion of financing channels mitigates financing constraints, offers financial support for innovation activities, stimulates greater R&D investments, and ultimately enhances enterprises’ innovation capabilities.

• Channel 3: Accumulating Human Capital

In comparison to the approval-based IPO system, the registration-based IPO system permits a more lenient examination of enterprise qualifications, thereby providing enterprises with more opportunities to gain access to the capital market. This, in turn, serves to expand the range of financing channels, alleviate financing constraints, and provide funds for enterprises to introduce external talent. Furthermore, compared with the approval-based IPO system, the registration-based IPO system can facilitate a reduction in information asymmetry, encourage shareholders to motivate and supervise the management, address the agency problem within enterprises, and encourage the management to proactively introduce external talent to invest in innovation activities [6]. In addition, the registration-based IPO system requires that IPO companies fully disclose key indicators related to scientific research and innovation as part of the listing conditions. This enables the external capital market to gain a more comprehensive understanding of enterprise innovation and raises the bar for the innovation abilities of enterprises. Consequently, the successful listing via the registration-based IPO system can indicate a certain level of enterprise innovation, which subsequently enhances the visibility and competitiveness of the enterprise, attracting external innovation talents into the enterprise to promote innovation.

Based on the above analysis, this paper proposes the following hypothesis:

Hypothesis 1 (H1).

Registration system reform can effectively promote the increase in R&D investment of IPO companies.

2.2.2. Registration System Reform, Listing Board and Enterprise Innovation

The registration system was implemented on the STAR Market and the ChiNext Market sequentially between 2019 and 2020. The STAR Market is distinguished by its focus on “hard technology”, positioning itself at the vanguard of global science and technology and aligning with the primary economic battleground and national priorities. The market primarily serves enterprises engaged in scientific and technological innovation aligned with national strategy, facilitates breakthroughs in key and core technologies, and is highly recognized by the market. The registration-based IPO system was introduced with the launch of the STAR Market on 13 June 2019. Up until 28 April 2020, the ChiNext Market implemented a reform of its registration system, thereby implementing an innovation-driven development strategy and positioning itself at the intersection of three and four innovations. The system was designed to serve the needs of growing innovative and entrepreneurial enterprises, with a particular emphasis on the innovation, creation, and creativity of issuers. It also aimed to facilitate the deep integration of traditional industries with new technologies, new industries, new forms of business, and new models [24].

The relevant provisions of the registration-based IPO system between the ChiNext Market and the STAR Market are largely consistent regarding issuance conditions, examination and registration procedures, issuance and underwriting, information disclosure, regulatory penalties, and other aspects. However, there are some discrepancies in technical details due to the differences in their respective positioning and development goals [26].

First, in terms of listing standards, the ChiNext Market imposes more rigorous requirements about the profitability of enterprises. In examining financial indicators, the STAR Market imposes additional requirements pertaining to technology and incorporates the “R&D investment” index to accommodate space for the listing of science and technology growth enterprises that encounter difficulties in generating profitable cash flow due to the high investment in early research and development. Furthermore, regarding the stability of the principal business and personnel, enterprises listed on the STAR Market are required to demonstrate that there have been no significant adverse changes in the main business and senior executives over the past two years. In addition, they must also provide evidence of the stability of their core technical personnel.

Second, in terms of the delisting system, the ChiNext Market is more rigorous than the STAR Market. Furthermore, the ChiNext Market has established more rigorous delisting criteria. For example, companies listed on the ChiNext Marke are subject to delisting if their trading market value falls below 500 million yuan for 20 consecutive days, whereas companies listed on the STAR Market are delisted if their trading market value drops below 300 million yuan over the same period. In terms of financial indicators, the latest financial and accounting reports must be accompanied by audit reports that are unable to express an opinion or a negative opinion. This is intended to reinforce the effectiveness of the clearing process. Additionally, the ChiNext Market has augmented its risk warning procedures and established a delisting risk warning system for financial, standard, and major illegal delisting.

Third, in terms of the trading system, the investor threshold of the ChiNext Market is more flexible than that of the STAR Market. In comparison to the STAR Market, the ChiNext Market has a more lenient threshold for the average daily assets in the securities and capital accounts over the 20 trading days preceding the submission of the application. The STAR Market requires a minimum of 500,000 yuan, whereas the ChiNext Market only necessitates a minimum of 100,000 yuan. Furthermore, the follow-up investment mechanism of the ChiNext Market is also more relaxed than that of the ChiNext Market. The registration system of the ChiNext Market only requires compulsory follow-up investment for four types of companies, including those that are unprofitable, those with a red-chip structure, those with special voting rights, and those that have been issued at a high price. In contrast, the STAR Market requires sponsor securities firms to use their funds to follow investment through their subsidiaries, with a proportion of 2% to 5% and a lock-up period of 2 years.

In light of the discrepancies in the implementation timeline of the registration-based IPO system between the STAR Market and the ChiNext Market, the variations in the specific provisions of the applicable policies, and the intrinsic dissimilarities in the scale, operational, and managerial modes of listed enterprises on the two boards, it is reasonable to hypothesize that the effects of the registration system reform on the enterprise innovation of IPO companies on the STAR Market and the ChiNext Market may be disparate. Thus, the following hypothesis is proposed:

Hypothesis 2 (H2).

The impact of registration system reform on the R&D investment of IPO companies in the STAR Market and the ChiNext Market exhibits significant differences.

3. Methodology

3.1. Data and Sample

The present study employs a sample comprising non-financial companies listed on the A-share market of the Shenzhen and Shanghai Stock Exchanges from 2019 to 2021 (The global market economy was significantly impacted by the COVID-19 pandemic between 2019 and 2021. Nevertheless, as all firms were situated within the same market environment, they experienced a similar degree of impact from the pandemic. Therefore, the Difference-in-Differences (DID) model, which employs a differential analysis between the treatment and control groups, effectively mitigates the influence of the pandemic on corporate innovation). Since its inception on 22 July 2019, The STAR Market has implemented a registration-based IPO system. In contrast, the ChiNext Market introduced a pilot IPO registration system on 24 August 2020. Therefore, the period from 2019 to 2023 represents a unique opportunity to examine the concurrent operation of the two systems. The impact of different listing systems on corporate innovation during the same period can be studied to avoid the influence of other macroeconomic factors. Furthermore, to accurately assess the impact of different listing systems on corporate innovation, it is essential to have a minimum of one year of observational data following the company’s initial public offering. In summary, the companies listed between 2019 and 2021 were selected as the research samples. Since the study requires data from three years before and after the listing of enterprises, the research sample commences in 2016 and concludes in 2022. Therefore, the actual period involved in the research sample is from 2016 to 2022.

The gradual implementation of the registration system reform creates a parallel sample period for the registration and approval system, allowing for the natural division of enterprises listed between 2019 and 2021 into two categories: registration system listing and approval system listing. This enables a quasi-natural experiment proposed in this paper to be conducted.

In terms of the experimental and control groups, all companies listed on the STAR Market and the ChiNext Market on or after 28 April 2020 are categorized as having been listed through the registration-based IPO system. Therefore, they are included in the experimental group. The IPO companies on the main board and those listed on the ChiNext Market prior to 27 April 2020 are categorized as belonging to the control group, as they were listed through the approval-based IPO system.

Table 1. Process of sample selection.

Screening Procedure	Number of Companies
Companies listed on A-shares from 1 January 2019 to 31 December 2021	1164
Excluded: sample companies listed on the Beijing Stock Exchange.	82
Excluded: sample companies whose issuance method is “proportional stock exchange” and “re-listing”.	6
Excluded: sample IPO companies in the financial industry.	24
Excluded: samples with missing values.	50
Companies that ultimately serve as the sample.	1002
Among: IPO companies from the STAR Market and the ChiNext Market.	715 (71.36%)
IPO companies from other boards.	287 (28.64%)

presents the screening process employed in the selection of the research samples discussed in this paper. The initial sample was constituted by A-share listed companies with listing time from 1 January 2019 to 31 December 2021, amounting to a total of 1164 companies. Subsequently, 82 companies listed on the Beijing Stock Exchange were excluded from the sample. Next, six sample companies whose issuance methods are “proportional stock exchange” or “re-listing were excluded from the analysis. Then, 24 listed companies operating within the financial industry were excluded. Finally, 50 samples with missing values were excluded from the analysis. The final sample consists of 1002 enterprises, of which 715 are listed on the STAR Market or on the ChiNext Market, accounting for 71.36% of the total sample. The total number of observations contributed by the sample companies over the observation period, spanning from 2016 to 2022, is 5244.

This study utilizes a sample of China’s A-share listed companies from the STAR Market, the ChiNext Market, and the main boards of the Shanghai and Shenzhen Stock Exchanges. The financial control variables, including company size (Size), asset-liability ratio (Lev), cash holding level (Cash), return on assets (ROA), and cash recovery rate of assets (CFO), are derived from the processing of financial data before and after IPO in the Wind database. Information regarding the composition of employees within the enterprises is also derived from the Wind database, while other data are obtained from the CSMAR database. To mitigate the influence of extreme values, this paper applies a winsorization technique to all continuous variables at the 1% and 99% quantiles.

3.2. Variable Definitions

3.2.1. Dependent Variables

The dependent variable in this study is enterprise innovation. Most extant studies employ a dual-dimensional approach to measuring enterprise innovation, encompassing both innovation input and innovation output. Given the lengthy cycle of innovation activities and the relatively brief implementation period of the registration system, there is an inherent limitation in the time series data when studying the impact of the registration system reform on enterprise innovation outputs. Accordingly, this paper employs R&D investment as a measure of enterprise innovation. Furthermore, in consideration of the influence of enterprise-scale on R&D investment in innovation, this paper employs the relative scale of R&D investment as a proxy variable for enterprise innovation, with the aim of examining the impact of listing on enterprise innovation in the context of the registration system reform. In particular, the dependent variable in this study is defined as the ratio of research and development (R&D) investment to operating revenue in the current year.

3.2.2. Independent Variables

The core independent variable of this study, RSI, serves as a proxy variable for listing through the registration-based IPO system. In instances where the sample observation is listed through the registration-based IPO system and belongs to the year of listing or the years after listing, the value of RSI is set to 1; otherwise, it is set to 0. Throughout the selected sample period, all companies listed on the STAR Market and the companies listed on the ChiNext Market after 28 April 2020, are categorized as listing through the registration-based IPO system. Conversely, IPO companies listed on the main board and those listed on the ChiNext Market before 27 April 2020, are categorized as listing through the approval-based IPO system.

The independent variable, RSI_IPO0, represents the dummy variables of the year of listing under the registration-based IPO system, while RSI_IPO_after represents the years after listing under the registration-based IPO system.

At the same time, RSI_IPO_after1, RSI_IPO_after2, and RSI_IPO_after3 are the dummy variables representing the one-, two-, and three-year periods after the initial listing under the registration-based IPO system, respectively.

3.2.3. Control Variables

In reference to the existing literature on enterprise innovation, this paper aims to control for several corporate finance and governance variables. First, the control variables of corporate finance include company size (Size), asset–liability ratio (Lev), cash holding level (Cash), return on assets (ROA), and cash recovery rate of assets (CFO). Second, in terms of governance variables, we control for board size (Board), the proportion of independent directors (Independent), and the shareholding ratio of the largest shareholder (First). The specific definitions of each control variable are presented in

Table 2. Definitions of variables.

Type	Symbol	Definition
Dependent Variable	R&D	It represents the ratio of R&D investment to operating revenue in the current year.
Independent Variables	RSI	It represents the sample firms listing under the registration-based IPO system. When the sample belongs to companies listed under the registration-based IPO system in the IPO year and subsequent years, the value is 1; otherwise, the value is 0.
	RSI_IPO0	It represents the sample companies in the year of listing under the registration-based IPO system. When the sample belongs to companies listed under the registration-based IPO system in the IPO year, the value is 1; otherwise, the value is 0.
	RSI_IPO_after	It represents the sample observed after the listing under the registration-based IPO system. When the sample belongs to companies listed under the registration-based IPO system in the year after listing, the value is 1; otherwise, the value is 0.
Control Variables	Size	It represents firm size, which is the natural logarithm of total assets at the end of the year.
	Lev	It represents asset–liability ratio, which is equal to the total liability divided by total assets at the end of the year.
	Cash	It represents cash holdings, which are equal to the monetary funds divided by total assets at the end of the year.
	ROA	It represents operating profitability, which is the net income in the current year divided by the total assets at the end of the year.
	CFO	It represents the asset cash recovery rate, which is equal to net operating cash flow divided by total assets.
	Board	It represents board size, which is equal to the number of board members at the end of the year.
	Independent	It represents the proportion of independent directors, which is equal to the proportion of the number of independent directors in the board of directors at the end of the year.
	First	It represents the shareholding ratio of the largest shareholder, which is equal to the proportion of the number of shares held by the largest shareholder to the total number of shares at the balance sheet date.
Mediating Variable	ATO	It represents total asset turnover, which is equal to operating income divided by the average balance of total assets.
Moderating Variables	SOE	It represents the nature of property rights, which takes the value of 1 when the samples belong to SOEs and the value of 0 when they belong to Non-SOEs.
	Labor	It represents the proportion of technical personnel, which is 1 when the proportion of technical personnel exceeds the average level of the sample in the year of listing, and 0 otherwise.

.

3.2.4. Mediating and Moderating Variables

To verify whether the registration system reform promotes enterprise innovation by addressing the agency problem, this paper introduces the mediating variable total asset turnover (ATO) as a measurement index of agency cost.

In terms of moderating variables, this paper introduces two dummy variables. First, to ascertain whether the alleviation of financial constraints acts as an important channel through which the registration system reform promotes enterprise innovation, the study introduces the property of ownership nature (SOE) as a proxy variable for financial constraints, in accordance with the methodology proposed by Zhang et al. [6]. Second, to ascertain whether the registration system reform promotes enterprise innovation through the accumulation of human capital, the study introduces the proportion of skilled personnel (Labor) as a proxy variable for innovation human capital. Furthermore, the samples are grouped according to the proportion of skilled personnel in the year of listing.

The precise definitions of each mediating and moderating variable are set out in Table 2.

3.3. Models

Enterprise innovation is affected by various factors, and if there are missing and unobservable variables related to the explanatory variables, this will result in a situation of endogeneity bias. To address this issue, the difference-in-differences (DID) model can be employed as a solution. This method addresses the issue of missing variables by first taking the first-order difference within each group and then calculating the difference between two groups that share a common trend. The IPO registration system reform can be regarded as a quasi-natural experiment, whereby enterprises listed through the registration-based IPO system and those listed through the approval-based IPO system are defined as the treatment group and the control group, respectively. According to the DID model, the impact of registration system reform on enterprise innovation can be evaluated by calculating the differences in R&D investment between the treatment group and the control group before and after the IPO.

Given that the registration system reform has been implemented in disparate sectors and phases since 2019, the treatment group is not simultaneously and uniformly affected by the policy impact. Therefore, to test Hypothesis 1, this paper constructs a staggered DID model as follows:

$${\mathrm{R}\&\mathrm{D}}_{\mathrm{i}\mathrm{t}}={{\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1\mathrm{R}\mathrm{S}\mathrm{I}}_{\mathrm{i}\mathrm{t}}+{\mathrm{C}\mathrm{O}\mathrm{N}\mathrm{T}\mathrm{R}\mathrm{O}\mathrm{L}\mathrm{S}}_{\mathrm{i}\mathrm{t}}+{\mathrm{Y}\mathrm{E}\mathrm{A}\mathrm{R}}_{\mathrm{t}}+{\mathrm{F}\mathrm{I}\mathrm{R}\mathrm{M}}_{\mathrm{i}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(1)

First, upon thorough validation, the sample data were found to satisfy the parallel trends assumption. Second, this paper is dedicated to a detailed examination of the immediate consequences of the reform of the registration system. At the same time, the policy change in question represents a clear temporal juncture. It is therefore proposed that the use of a linear model will facilitate a more efficacious and accurate delineation of causal relationships.

In Model (1), the dependent variable is ${\mathrm{R}\&\mathrm{D}}_{\mathrm{i}\mathrm{t}}$, while ${\mathrm{R}\mathrm{S}\mathrm{I}}_{\mathrm{i}\mathrm{t}}$ is the core independent variable. The regression coefficient of ${\mathrm{R}\mathrm{S}\mathrm{I}}_{\mathrm{i}\mathrm{t}}$ quantifies the incremental impact of listing through the registration-based IPO system on enterprise innovation compared with listing through the approval-based IPO system. This effect can be attributed to the registration system reform and its influence on innovation. CONTROLS represents the control variables, including company size (Size), asset–liability ratio (Lev), cash holding level (Cash), return on assets (ROA), cash recovery rate of assets (CFO), board size (Board), the proportion of independent directors (Independent) and the shareholding ratio of the largest shareholder (First). YEAR_t denotes time-fixed effects, while FIRM_i denotes individual fixed effects.

To further examine the dynamic impact of the registration system reform on enterprise innovation, this paper extends Model (1) to develop the following empirical models:

$${\mathrm{R}\&\mathrm{D}}_{\mathrm{i}\mathrm{t}}={{\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1\mathrm{R}\mathrm{S}\mathrm{I}\_\mathrm{I}\mathrm{P}\mathrm{O}0}_{\mathrm{i}\mathrm{t}}+{{\mathsf{\alpha }}_2\mathrm{R}\mathrm{S}\mathrm{I}\_\mathrm{I}\mathrm{P}\mathrm{O}\_\mathrm{a}\mathrm{f}\mathrm{t}\mathrm{e}\mathrm{r}}_{\mathrm{i}\mathrm{t}}+{\mathrm{C}\mathrm{O}\mathrm{N}\mathrm{T}\mathrm{R}\mathrm{O}\mathrm{L}\mathrm{S}}_{\mathrm{i}\mathrm{t}}+{\mathrm{Y}\mathrm{E}\mathrm{A}\mathrm{R}}_{\mathrm{t}}+{\mathrm{F}\mathrm{I}\mathrm{R}\mathrm{M}}_{\mathrm{i}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(2)

$$\begin{gathered} {\mathrm{R}\&\mathrm{D}}_{\mathrm{i}\mathrm{t}}={{\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1\mathrm{R}\mathrm{S}\mathrm{I}\_\mathrm{I}\mathrm{P}\mathrm{O}0}_{\mathrm{i}\mathrm{t}}+{{\mathsf{\alpha }}_2\mathrm{R}\mathrm{S}\mathrm{I}\_\mathrm{I}\mathrm{P}\mathrm{O}\_\mathrm{a}\mathrm{f}\mathrm{t}\mathrm{e}\mathrm{r}1}_{\mathrm{i}\mathrm{t}}+ \\ {{\mathsf{\alpha }}_3\mathrm{R}\mathrm{S}\mathrm{I}\_\mathrm{I}\mathrm{P}\mathrm{O}\_\mathrm{a}\mathrm{f}\mathrm{t}\mathrm{e}\mathrm{r}2}_{\mathrm{i}\mathrm{t}}{+{{\mathsf{\alpha }}_4\mathrm{R}\mathrm{S}\mathrm{I}\_\mathrm{I}\mathrm{P}\mathrm{O}\_\mathrm{a}\mathrm{f}\mathrm{t}\mathrm{e}\mathrm{r}3}_{\mathrm{i}\mathrm{t}}+\mathrm{C}\mathrm{O}\mathrm{N}\mathrm{T}\mathrm{R}\mathrm{O}\mathrm{L}\mathrm{S}}_{\mathrm{i}\mathrm{t}}+{\mathrm{Y}\mathrm{E}\mathrm{A}\mathrm{R}}_{\mathrm{t}}+{\mathrm{F}\mathrm{I}\mathrm{R}\mathrm{M}}_{\mathrm{i}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}} \end{gathered}$$

(3)

Model (2) is used to ascertain whether there is a discernible difference in the impact of the registration system reform on innovation during the year of listing and in the three years after listing. Model (3) is used to investigate the evolving effect of the registration system reform on innovation in the first, second and third years after listing.

When testing the overall impact of the registration system reform on enterprise innovation (H1), this study uses the full sample to estimate Models (1), (2), and (3), respectively. When comparing the differences in the impact of the registration system reform on enterprise innovation between the STAR Market and the ChiNext Market (H2), the study initially uses the listed companies on the ChiNext Market to estimate Models (1), (2) and (3), respectively, with the objective of testing the impact of registration system reform of the ChiNext Market on enterprise innovation. Subsequently, the study uses the listed companies on the STAR Market and those on the main boards to estimate Models (1), (2), and (3), respectively, with the objective of examining the impact of registration system reform of the STAR Market on enterprise innovation.

4. Results

4.1. Descriptive Analysis

Table 3. Descriptive statistics of the whole sample.

Variable	N	Mean	Median	S.D.	Min	Max
R&D	5244	0.0773	0.0491	0.0866	0.0009	0.5731
RSI	5244	0.6310	1.0000	0.4827	0.0000	1.0000
Size	5244	11.8807	11.8000	1.0126	9.8326	15.3005
Lev	5244	0.3346	0.3129	0.1770	0.0451	0.8035
Cash	5244	0.0523	0.0293	0.1410	−0.3050	0.5560
ROA	5244	0.0884	0.0798	0.0726	−0.1737	0.3278
CFO	5244	0.3448	0.2461	0.3458	−0.1428	1.8990
Board	5244	1.8714	1.9459	0.3906	0.6931	2.5649
Independent	5244	0.3778	0.3750	0.0888	0.1429	0.6667
First	5244	40.6681	38.2264	17.5688	11.3500	88.7196

presents the descriptive statistics of the main variables under investigation in this study. The results indicate that the mean value of R&D is 0.0773, suggesting that the mean ratio of R&D investment to operating revenue for the sample companies is approximately 7.73%. The maximum and minimum values of R&D are 0.5731 and 0.0009, respectively, indicating a significant disparity in R&D investment among the samples. The median value of R&D is 0.0491, indicating that most sample enterprises allocate a relatively minor proportion of operating revenue to R&D investment. The mean value of RSI is 0.631, indicating that companies listed through the registration-based IPO system account for 63.1% of the entire sample.

Table 4. Comparisons between the treatment group and the control group.

Variable	(1) RSI = 0, N = 1937		(2) RSI = 1, N = 3307		Diff. = (2) − (1)
	Mean1	Median1	Mean2	Median2	Mean Diff.	Median Diff.
R&D	0.0460	0.0370	0.0960	0.0600	0.0490 ***	0.0230 ***
Size	12.1400	12.0500	11.7300	11.6600	−0.4130 ***	−0.3900 ***
Lev	0.3550	0.3400	0.3230	0.2950	−0.0310 ***	−0.0450 ***
Cash	0.0450	0.0290	0.0570	0.0300	0.0120 ***	0.0010
ROA	0.0930	0.0830	0.0860	0.0780	−0.0070 ***	−0.0050 ***
CFO	0.4290	0.3100	0.2950	0.2190	−0.1340 ***	−0.0910 ***
Board	1.8300	1.9460	1.8960	1.9460	0.0660 ***	0.0000 ***
Independent	0.3810	0.3750	0.3760	0.3750	−0.0060 **	0.0000 ***
First	44.1000	41.4000	38.6600	36.6000	−5.4430 ***	−4.8000 ***

Note. The symbols *** and ** indicate significance at the 1% and 5% levels, respectively.

presents a comparison between the treatment group (RSI = 1), which consists of companies listed through the registration-based IPO system, and the control group (RSI = 0), which consists of companies listed through the approval-based IPO system. The analysis examines the differences in the main variables between the two groups. The results indicate that, based on the test of mean difference, the mean values of the enterprise innovation variable (R&D) in the treatment group and control group are 0.0960 and 0.0460, respectively. The difference between the two groups is significant at the 1% level. In the median difference test, the median of R&D in the treatment group and control group are 0.0600 and 0.0370, respectively. The difference between the two groups is 0.0230 and significant at the 1% level. These findings suggest that companies listed through the registration-based IPO system exhibit a higher level of innovation investment compared to those listed through the approval-based IPO system. This provides initial evidence supporting Hypothesis 1. Furthermore, in terms of the differences in company characteristic variables, the companies listed through the registration-based IPO system tend to have a smaller size, a lower debt level, poorer operating performance and financial performance, but a higher cash holding level, in comparison to the samples listing through the approval-based IPO system. In terms of the differences in governance structure, the size of the board of directors is larger in companies listed through the registration-based IPO system than in those listed through the approval-based IPO system. However, the proportion of independent directors and ownership concentration is lower in the former.

4.2. Correlation Analysis

Table 5. Pearson correlations between main variables.

Variable	R&D	RSI	Size	Lev	ROA	CFO	Cash	Board
RSI	0.07930 ***	1.0000
Size	−0.0600 ***	0.4984 ***	1.0000
Lev	−0.2306 ***	−0.3174 ***	0.1856 ***	1.0000
ROA	−0.3282 ***	−0.3727 ***	−0.3557 ***	−0.1529 ***	1.0000
CFO	−0.1794 ***	0.5642 ***	0.3513 ***	−0.0816 ***	−0.0297 **	1.0000
Cash	−0.0027	0.0583 ***	−0.0162	−0.1393 ***	0.1275 ***	0.0439 **	1.0000
Board	0.0823 ***	0.3256 ***	0.1980 ***	−0.0751 ***	−0.1929 ***	0.1944 ***	−0.0479 ***	1.0000
Independent	0.0308 **	0.0328 **	0.0511 ***	−0.0146	0.0114	0.0126	−0.0045	−0.2011 ***
First	−0.2584 ***	−0.2214 ***	−0.0076	0.1053 ***	0.2111 ***	−0.0389 **	0.0046	−0.1990 ***

Note. The symbols *** and ** indicate significance at the 1% and 5% levels, respectively.

presents the Pearson correlations between the primary variables under investigation in this study. The table shows that the correlation between the dependent variable (R&D) and the independent variable (RSI) is significantly positive at the 1% level, indicating that companies listed through the registration-based IPO system invest more in R&D than companies listed through the approval-based IPO system. This finding initially supports Hypothesis 1. The results also suggest that the dependent variable (R&D) is negatively related to the company size (Size), asset–liability ratio (Lev), profitability (CFO), return on assets (ROA), and ownership concentration (First), but positively related to the board size (Board) and proportion of independent directors (Independent). In addition, the absolute values of the correlations between the respective variables are all below 0.5, indicating that the empirical models constructed in this paper do not exhibit significant multicollinearity issues.

4.3. Multiple Regression Analysis

4.3.1. The Impact of Registration System Reform on Enterprise Innovation

Table 6. Regression results of the impact of registration system reform on enterprise innovation.

Variable	Model (1): R&D	Model (2): R&D	Model (3): R&D
RSIit	0.0059 ***
	(2.71)
RSI_IPO0		0.0006	0.0008
		(0.26)	(0.36)
RSI_IPO_after		0.0111 ***
		(4.18)
RSI_IPO_after1			0.0092 ***
			(3.58)
RSI_IPO_after2			0.0196 ***
			(4.86)
RSI_IPO_after3			0.0332 ***
			(4.53)
Size	−0.0206 ***	−0.0206 ***	−0.0206 ***
	(−6.99)	(−7.01)	(−7.05)
Lev	−0.0165 **	−0.0201 **	−0.0246 ***
	(−2.12)	(−2.55)	(−3.04)
ROA	−0.2628 ***	−0.2679 ***	−0.2689 ***
	(−9.35)	(−9.49)	(−9.57)
CFO	−0.0096 ***	−0.0087 ***	−0.0077 ***
	(−4.61)	(−4.22)	(−3.73)
Cash	−0.0188 ***	−0.0122 ***	−0.0134 ***
	(−5.37)	(−3.33)	(−3.65)
Board	0.0058 **	0.0051 *	0.0050 *
	(1.97)	(1.75)	(1.70)
First	0.0004 *	0.0004 **	0.0003
	(1.79)	(1.97)	(1.41)
Independent	−0.0063	−0.0056	−0.0035
	(−0.55)	(−0.49)	(−0.30)
Constant	0.3127 ***	0.3137 ***	0.3219 ***
	(8.83)	(8.90)	(9.05)
Firm FE	YES	YES	YES
Year FE	YES	YES	YES
R-squared	0.257	0.263	0.270
F	19.43 ***	18.28 ***	17.52 ***
Observations	5244	5244	5244

Note. The symbols ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively. The values in parentheses represent the corresponding t-test values.

reports the regression results pertaining to the impact of the registration system reform on enterprise innovation. The estimated results of Model (1) show that the coefficient of RSI_it is 0.0059, which is statistically significant at the 1% level. This suggests that compared with the approval-based IPO system, the registration-based IPO system has a significant incremental positive impact on enterprise innovation. In terms of economic implications, the promotion effect of the registration-based IPO system on the intensity of research and development is 0.59% higher than that of the approval-based IPO system. The above results indicate that, in general, the registration system reform has a significant promotion effect on the R&D investment of IPO companies, thereby providing substantial support for Hypothesis 1. Furthermore, the estimation results of Model (2) demonstrate that the promotion effect of the registration system reform on enterprise innovation is not significant in the year of listing. However, it becomes significant after listing, indicating that the impact of the registration system reform on enterprise innovation is reflected after listing. This implies that the promotion effect has a lag. Additionally, the estimation results of Model (3) indicate that the promotion effect of the registration system reform on enterprise innovation is significant and increases progressively over the three years following the listing. This suggests that the impact of the registration system reform on enterprise innovation exhibits a cumulative effect.

4.3.2. Differences in the Impact of Registration System Reform on Enterprise Innovation between the STAR Market and the ChiNext Market

Table 7. Differences in the impact of the registration system reform on enterprise innovation between the STAR Market and the ChiNext Market.

R&D	Model (1)		Model (2)		Model (3)
	ChiNext Market	STAR Market	ChiNext Market	STAR Market	ChiNext Market	STAR Market
RSIit	−0.0068 **	0.0123 ***
	(−2.12)	(4.09)
RSI_IPO0			−0.0068 **	0.0042	−0.0067 **	0.0043
			(−2.59)	(1.41)	(−2.59)	(1.46)
RSI_IPO_after			−0.0069	0.0190 ***
			(−1.58)	(5.31)
RSI_IPO_after1					−0.0068	0.0158 ***
					(−1.59)	(4.56)
RSI_IPO_after2					−0.0074	0.0263 ***
					(−1.30)	(5.38)
RSI_IPO_after3						0.0355 ***
						(4.81)
Size	−0.0156 ***	−0.0223 ***	−0.0156 ***	−0.0223 ***	−0.0156 ***	−0.0222 ***
	(−3.16)	(−6.53)	(−3.15)	(−6.55)	(−3.15)	(−6.55)
Lev	−0.0189 **	−0.0201 **	−0.0189 **	−0.0254 ***	−0.0188 **	−0.0295 ***
	(−2.37)	(−2.11)	(−2.38)	(−2.61)	(−2.35)	(−2.96)
ROA	−0.1837 ***	−0.3001 ***	−0.1837 ***	−0.3074 ***	−0.1837 ***	−0.3074 ***
	(−4.09)	(−9.27)	(−4.10)	(−9.46)	(−4.10)	(−9.50)
CFO	−0.0053 **	−0.0101 ***	−0.0053 **	−0.0089 ***	−0.0053 **	−0.0080 ***
	(−2.52)	(−3.78)	(−2.52)	(−3.34)	(−2.51)	(−3.03)
Cash	−0.0131 **	−0.0212 ***	−0.0131 **	−0.0129 ***	−0.0131 **	−0.0142 ***
	(−2.43)	(−4.82)	(−2.26)	(−2.85)	(−2.27)	(−3.10)
Board	−0.0044	0.0078 **	−0.0044	0.0069 **	−0.0044	0.0070 **
	(−1.25)	(2.34)	(−1.25)	(2.08)	(−1.24)	(2.11)
First	−0.0000	0.0004	−0.0000	0.0004	−0.0000	0.0003
	(−0.19)	(1.49)	(−0.20)	(1.57)	(−0.19)	(1.19)
Independent	0.0008	−0.0025	0.0008	−0.0008	0.0008	0.0010
	(0.07)	(−0.18)	(0.07)	(−0.05)	(0.07)	(0.07)
Constant	0.2568 ***	0.3428 ***	0.2569 ***	0.3449 ***	0.2568 ***	0.3495 ***
	(4.25)	(8.21)	(4.25)	(8.32)	(4.24)	(8.40)
Firm FE	YES	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES	YES
R-squared	0.209	0.294	0.209	0.302	0.210	0.308
F	8.97 ***	17.17 ***	8.82 ***	16.41 ***	8.32 ***	15.51 ***
Observations	1758	3947	1758	3947	1758	3947

Note. The symbols *** and ** indicate significance at the 1% and 5% levels, respectively. The values in parentheses represent the corresponding t-test values. As the registration system of the ChiNext Market has been implemented since 2020, the effect in the third year after listing cannot be tested temporarily.

presents a comparative analysis of the impact of the registration system reform on enterprise innovation between the STAR Market and the ChiNext Market. In examining the impact of the ChiNext Market registration system reform on enterprise innovation, the treatment group consists of companies listed on the ChiNext Market under the registration-based IPO system, while the control group consists of companies listed on the ChiNext Market under the approval-based IPO system. In tests of the impact of the STAR Market registration system reform on enterprise innovation, the treatment group comprises companies listed on the STAR Market, while the control group consists of companies listed on the main board.

The estimated results of Model (1) for the ChiNext Market show that the coefficient of RSI_it is −0.0068, which is statistically significant at the 5% level. This suggests that the level of R&D investment in companies listed on the ChiNext Market under the registration-based IPO system is significantly lower in comparison to companies listed through the approval-based IPO system following listing. In other words, the reform of the registration system has a significant inhibitory effect on enterprise innovation for companies listed on the ChiNext Market. However, in the regression with enterprises on the STAR Market as the treatment group, the coefficient of RSI_it is 0.0123, which is statistically significant at the 1% level. This indicates that, in comparison to companies listed through the approval-based IPO system, companies listed through the registration-based IPO system can significantly promote the innovation of enterprises on the STAR Market. It can therefore be concluded that the effect of registration system reform on enterprise innovation differs significantly between IPO companies on the ChiNext Market and IPO companies on the STAR Market, thereby supporting the hypothesis that the reform has a differential effect on different markets.

The estimated results of Model (2) indicate that the inhibitory effect of the registration system reform on the R&D intensity is significant in the year of listing for companies on the ChiNext Market, but becomes insignificant after listing. Moreover, the promotion effect of the registration system reform on the R&D intensity is not significant in the year of listing for companies on the STAR Market, but becomes significant after listing.

The estimation results of Model (3) suggest that the ChiNext Market registration system reform exerts an inhibitory effect on the R&D intensity exclusively during the year in which companies are listed. However, the promotion effect of the STAR Market registration system reform on the R&D intensity is only significant after listing, with an increase over time after listing.

In conclusion, the reform of the registration system can indeed facilitate enterprise innovation. This promotion effect is observed exclusively in IPO companies on the STAR Market, and it is evidenced by a lag and cumulative effect over time following the initial listing.

4.3.3. Parallel Trend Tests

A fundamental prerequisite for employing the DID model is to ensure compliance with the parallel trend assumption, which postulates the existence of a consistent trajectory in the level of enterprise R&D investment between the treatment group and the control group prior to the intervention. Failure to meet the parallel trends assumption may result in the generation of biased estimation results for the DID model. To assess the conformity of the DID model with the parallel trend hypothesis, this study employs two common methods, which have been influenced by existing research.

The first approach is to conduct a qualitative examination through the utilization of parallel trend graphs. Should the change trend of the dependent variable between the treatment group and the control group before the intervention be wholly or largely consistent, this would indicate that the model is in accordance with the hypothesis of parallel trends. In this study, the treatment group comprises IPO companies listed through the registration-based IPO system, while the control group consists of IPO companies listed through the approval-based IPO system. In accordance with the methodology proposed by Beck et al. [40], this paper selects the listing year of each enterprise as the base period for conducting a parallel trend test. Specifically, the investigation seeks to ascertain whether the evolving trajectory of the mean value of R&D investment in the treatment group and the control group over the three years preceding listing is consistent. Figure 1 illustrates that during the three-year period prior to listing, the treatment group exhibited higher levels of R&D investment in comparison to the control group. However, the changing trend of the two groups was found to be generally similar, with no significant fluctuations or obvious increasing or decreasing trends.

The second method is to use quantitative analysis. In accordance with the methodology proposed by Amore et al. [41], this study adopts the event study approach to develop the following model for testing parallel trends:

$${\mathrm{R}\&\mathrm{D}}_{\mathrm{i}\mathrm{t}}=\mathsf{\alpha }+{\sum }_{\mathrm{n}=1}^2{\mathsf{\beta }}_{\mathrm{n}}{\mathrm{P}\mathrm{r}\mathrm{e}}_{\mathrm{i}\mathrm{t}}^{\mathrm{n}}+{\sum }_{\mathrm{n}=0}^3{\mathsf{\theta }}_{\mathrm{n}}{\mathrm{P}\mathrm{o}\mathrm{s}\mathrm{t}}_{\mathrm{i}\mathrm{t}}^{\mathrm{n}}+{\mathrm{C}\mathrm{O}\mathrm{N}\mathrm{T}\mathrm{R}\mathrm{O}\mathrm{L}\mathrm{S}}_{\mathrm{i}\mathrm{t}}+{\mathrm{Y}\mathrm{E}\mathrm{A}\mathrm{R}}_{\mathrm{t}}+{\mathrm{F}\mathrm{I}\mathrm{R}\mathrm{M}}_{\mathrm{i}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(4)

In the model, the dependent variable (R&D) represents R&D investment. For the independent variables, Preⁿ represents a dummy variable of n years before the listing of the registration-based IPO system, whereas Postⁿ represents a dummy variable of n years after the listing of the registration-based IPO system. CONTROLS represents control variables, YEAR represents the time-fixed effects, and FIRM represents the individual fixed effects. These variables are consistent with those used in Model (1).

Table 8. Results of the parallel trend test of the multi-time DID model.

Variables	R&D
${\mathrm{P}\mathrm{r}\mathrm{e}}_{\mathrm{i}\mathrm{t}}^2$	−0.0019
	(−1.18)
${\mathrm{P}\mathrm{r}\mathrm{e}}_{\mathrm{i}\mathrm{t}}^1$	−0.0028
	(−1.27)
${\mathrm{P}\mathrm{o}\mathrm{s}\mathrm{t}}_{\mathrm{i}\mathrm{t}}^0$	−0.0012
	(−0.48)
${\mathrm{P}\mathrm{o}\mathrm{s}\mathrm{t}}_{\mathrm{i}\mathrm{t}}^1$	0.0072 **
	(2.49)
${\mathrm{P}\mathrm{o}\mathrm{s}\mathrm{t}}_{\mathrm{i}\mathrm{t}}^2$	0.0173 ***
	(4.02)
${\mathrm{P}\mathrm{o}\mathrm{s}\mathrm{t}}_{\mathrm{i}\mathrm{t}}^3$	0.0305 ***
	(4.11)
CONTROLS	YES
Firm FE	YES
Year FE	YES
R-squared	0.270
F	16.59 ***
Observations	5244

Note. The symbols *** and ** indicate significance at the 1% and 5% levels, respectively. The values in parentheses represent the corresponding t-test values. To conserve space, the regression results of control variables and constant terms are omitted in the table.

presents the results of the parallel trend test for the staggered difference-in-differences (DID) model. As can be observed from the table, the coefficient estimates of the dummy variables for the year of the registration-based IPO system listing and the first two years before listing are not statistically significant. This indicates that there is no significant difference in the R&D investment intensity between the experimental group and the control group before the listing, thereby supporting the parallel trend hypothesis of the DID model. However, the estimated coefficients for the three years after the registration system listing are all statistically significant, indicating that the impact of the registration-based IPO system listing and the approval-based IPO system listing on enterprise innovation is significantly different. This suggests that the registration system reform has a policy effect. It is noteworthy that the estimated regression coefficients for Postⁿ are positive, indicating that each period within the three years after the registration system listing has a significantly positive impact on corporate R&D investment. This finding serves to reinforce the conclusion that the reform of the registration system is conducive to enterprise innovation, which is in alignment with the preceding analysis results.

4.3.4. Robustness Tests

To confirm that the changes in enterprise innovation performance are indeed caused by listing under the registration system rather than by other systematic differences during the observation period, this study employs the propensity score matching (PSM) method to conduct the robustness test. The samples are divided into the treatment group (RSI = 1) and the control group (RSI = 0) based on whether they are listed under the registration-based IPO system. The propensity scores for the two groups are calculated using all the control variables in the main regression model (1) with the logit model, and then the nearest neighbor is matched 1:1 with the propensity score. Finally, 4452 valid observation samples are retained.

Table 9. Regression results based on the propensity score matched sample.

R&D	Full Sample		ChiNext Market Registration System		STAR Market Registration System
	Model (2)	Model (3)	Model (2)	Model (3)	Model (2)	Model (3)
RSI_IPO0	−0.0018	−0.0010	−0.0076 **	−0.0076 **	0.0011	0.0013
	(−0.72)	(−0.42)	(−2.34)	(−2.34)	(0.32)	(0.40)
RSI_IPO_after	0.0084 ***		−0.0084		0.0155 ***
	(2.72)		(−1.50)		(3.95)
RSI_IPO_after1		0.0078 **		−0.0084		0.0133 ***
		(2.56)		(−1.50)		(3.45)
RSI_IPO_after2		0.0187 ***		−0.0093		0.0240 ***
		(4.17)		(−1.28)		(4.61)
RSI_IPO_after3		0.0323 ***				0.0328 ***
		(4.09)				(4.14)
CONTROLS	YES	YES	YES	YES	YES	YES
Firm FE	YES	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES	YES
R-squared	0.265	0.273	0.205	0.205	0.314	0.320
Observations	4452	4452	1601	1601	3156	3156

Note. The symbols *** and ** indicate significance at the 1% and 5% levels, respectively. The values in parentheses represent the corresponding t-test values. To conserve space, the regression results of control variables and constant terms are omitted in the table. As the registration system of the ChiNext Market has been implemented since 2020, the effect in the third year after listing cannot be tested temporarily.

reports the regression results based on the propensity score matched sample.

The estimation results for models (2) and (3) based on the full sample after matching are displayed in Columns 1 and 2 of Table 9. The results indicate that the coefficients of RSI_IPO0 are not statistically significant. However, when considering the overall situation in the three years after listing and the specific performance of each year, the estimated coefficients are significantly positive, with a gradual increase observed for each year after listing. The above results suggest that, in comparison to listing through the approval-based IPO system, listing through the registration-based IPO system exhibits a significant incremental positive impact on enterprise innovation. Consequently, the reform of the registration system markedly encourages the R&D investment of IPO companies, exhibiting a cumulative and lagging effect. The above conclusions are consistent with the analysis results presented in Table 6, thereby providing further support for Hypothesis 1.

Columns 3 to 6 of Table 9 report the regression results when further distinctions are made between the sectors of the matched sample. It can be observed that in the ChiNext Market samples, the estimated coefficient of RSI_IPO0 is significantly negative at the 5% level. In contrast, the estimated coefficients for the three years after listing and each year after listing are not statistically significant, indicating that the registration system reform only has a significantly negative impact on the R&D investment intensity of the enterprises on the ChiNext Market in the year of listing. In the samples of IPO companies on the STAR Market and companies listed under the approval-based IPO system, the estimated coefficients of RSI_IPO0 are not statistically significant. However, the estimated coefficients of RSI_IPO_after and RSI_IPO_after* (RSI_IPO_after1, RSI_IPO_after2, RSI_IPO_after3) are significantly positive at the 1% level, with estimated coefficients for each year after listing exhibiting a gradual increase. These findings indicate that for IPO companies on the STAR Market, the registration system reform has a significant promotion effect on enterprise R&D investment, with a certain lag and cumulative effect. In conclusion, the impact of the registration system reform on R&D investment differs significantly between IPO companies on the ChiNext Market and IPO companies on the STAR Market, which is consistent with the analysis results in Table 7 and further supports Hypothesis 2.

4.3.5. Mechanism Tests

• Channel 1: Reducing Agency Costs

To verify whether the registration system reform promotes enterprise innovation by addressing the agency problem, this paper introduces the mediating variable total asset turnover as the measurement index of agency costs. Furthermore, the two-step method, as proposed by Jiang [42], is employed to test the mediating effect of agency costs.

The second-step empirical model is constructed as follows:

$${\mathrm{A}\mathrm{T}\mathrm{O}}_{\mathrm{i}\mathrm{t}}={{\mathsf{\beta }}_0+{\mathsf{\beta }}_1\mathrm{R}\mathrm{S}\mathrm{I}\_\mathrm{I}\mathrm{P}\mathrm{O}0}_{\mathrm{i}\mathrm{t}}{{+{\mathsf{\beta }}_2\mathrm{R}\mathrm{S}\mathrm{I}\_\mathrm{I}\mathrm{P}\mathrm{O}\_\mathrm{a}\mathrm{f}\mathrm{t}\mathrm{e}\mathrm{r}}_{\mathrm{i}\mathrm{t}}+\mathrm{C}\mathrm{O}\mathrm{N}\mathrm{T}\mathrm{R}\mathrm{O}\mathrm{L}\mathrm{S}}_{\mathrm{i}\mathrm{t}}+{\mathrm{Y}\mathrm{E}\mathrm{A}\mathrm{R}}_{\mathrm{t}}+{\mathrm{F}\mathrm{I}\mathrm{R}\mathrm{M}}_{\mathrm{i}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(5)

Among the variables under consideration, the mediating variable, ATO, represents total asset turnover. This is calculated by dividing the operating income by the average balance of total assets. A higher value of this indicator indicates lower agency costs.

Table 10. Regression results of the mechanism test of the impact of the registration system reform on enterprise innovation.

Variables	(1) Reduce Agency Costs	(2) Alleviating Financing Constraints		(3) Accumulating Human Capital
	Full Sample	SOE = 1	SOE = 0	Labor = 1	Labor = 0
RSI_IPO0	0.0577 ***	0.0020	0.0006	−0.0039	0.0006
	(4.84)	(0.37)	(0.27)	(−0.94)	(0.39)
RSI_IPO_after	0.0326 **	0.0045	0.0118 ***	0.0073	0.0051 **
	(2.32)	(0.92)	(4.15)	(1.48)	(2.19)
CONTROLS	YES	YES	YES	YES	YES
Firm FE	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES
R-squared	0.699	0.112	0.271	0.308	0.197
F value	224.25 ***	2.28 ***	18.12 ***	16.01 ***	7.59 ***
Observations	5244	428	4816	2357	2153

Note. The symbols *** and ** indicate significance at the 1% and 5% levels, respectively. The values in parentheses represent the corresponding t-test values. To conserve space, the regression results of control variables and constant terms are omitted in the table.

, column (1) presents the estimated results obtained through the application of Model (5) to assess the impact of the registration system reform on the agency costs of enterprises. The results suggest that the regression coefficients of RSI_IPO0 and RSI_IPO_after are both significantly positive, indicating that the registration system reform can effectively improve the total asset turnover of enterprises and mitigate the agency problem within firms. As previously stated, the mitigation of agency problems can, to a certain extent, alleviate the current situation whereby agents excessively focus on short-term performance and ignore innovation. This is because the registration system reform can encourage agents to make decisions conducive to innovation activities, which is beneficial to shareholders. Therefore, the registration system reform can promote enterprise innovation by reducing agency costs.

• Channel 2: Alleviate financing constraints

The differences in capital structures among diverse enterprises prior to listing will consequently give rise to disparate degrees of financing constraints encountered by these enterprises [43]. The extent to which the registration system reform alleviates financing constraints also varies. The ownership of state-owned enterprises (SOEs) enables them to obtain additional funds, including government subsidies, and access wider financing channels than non-state-owned enterprises (non-SOEs). At the same time, the involvement of government credit can serve to reinforce the capacity of enterprises to obtain commercial credit and fortify the bank–enterprise relationship, thereby facilitating the use of loans by state-owned enterprises [44]. Therefore, the financial constraints faced by non-state-owned enterprises prior to listing are more pronounced than those encountered by state-owned enterprises. Consequently, the alleviation of financial constraints resulting from the registration system reform will be more evident in non-state-owned enterprises, given the more relaxed listing conditions compared to the approval-based IPO system.

To verify whether the alleviation of financial constraints serves as an important channel through which the registration system reform promotes enterprise innovation, the study introduces the property of ownership nature (SOE) as a proxy variable for financial constraints. Following Zhang et al. [6], we use the samples of state-owned enterprises and non-state-owned enterprises to estimate Model (1).

Table 10, column (2), presents the estimation results of Model (2) by using samples of state-owned enterprises and non-state-owned enterprises, respectively. In the year of listing under the registration-based IPO system, the coefficients of RSI_IPO0 are both positive and insignificant for state-owned enterprises and non-state-owned enterprises. This finding is consistent with the test results of H1, indicating that the promotion effect of the registration system reform on enterprise innovation cannot be reflected in the year of listing for both state-owned enterprises and non-state-owned enterprises. However, following the listing, the coefficient of RSI_IPO_after in the regression results of state-owned enterprises is positive but not significant, while the coefficient of RSI_IPO_after in the regression results of non-state-owned enterprises is significantly positive at the 1%level. This indicates that the impact of the registration system reform on the R&D investment intensity differs significantly between state-owned enterprises and non-state-owned enterprises. Furthermore, in the non-state-owned enterprise sample with a higher degree of financial constraints, the registration system reform has a more pronounced effect on enterprise innovation. This is evidenced by the fact that, in the three years following listing, the reform is found to have a significant positive impact on innovation, particularly in enterprises facing greater financial constraints. In conclusion, the registration system reform can promote enterprise innovation by alleviating financial constraints.

• Channel 3: Accumulating Human Capital

Given the disparate human capital conditions of various enterprises prior to listing, it is evident that the impact of registration system reform on the attraction of external talent also varies. The contribution of technical personnel to enterprise research and development activities is of great consequence, so enterprises with a low proportion of such personnel face a greater scarcity of innovation talents. This makes the effect of registration system reform on attracting external talents more pronounced for them. If the registration system reform can further attract external talents to promote enterprise innovation, its impact on enterprises with a low proportion of technical personnel at the time of listing will be particularly pronounced.

To verify whether the registration system reform promotes enterprise innovation through the accumulation of human capital, the study introduces the proportion of skilled personnel (Labor) as a proxy variable for innovation human capital, and groups the samples according to the proportion of skilled personnel in the year of listing. On this basis, Model (1) is estimated by grouped regression.

Table 10, column 3, presents the estimation results of Model (2), employing samples from two groups: one comprising a relatively high proportion of skilled personnel and the other comprising a relatively low proportion of such personnel. The results indicate that following the initial public offering (IPO) through the registration-based system, the coefficient of RSI_IPO_after for enterprises with a high proportion of technical personnel is positive but not statistically significant, while the coefficient of RSI_IPO_after for enterprises with a low proportion of technical personnel is significantly positive at the 5% level. It can thus be concluded that the impact of the registration system reform on enterprise innovation differs significantly depending on the proportion of technical personnel in the enterprise in question. Moreover, for enterprises with low proportions of technical personnel at the time of listing, the registration system reform has a more pronounced effect on promoting R&D investment, thereby substantiating the channel of human capital accumulation. Furthermore, the registration system reform is shown to promote enterprise innovation through the accumulation of human capital over the three-year period following listing.

5. Discussion and Conclusions

The reform of the registration system represents a significant institutional reform within China’s capital market. This paper employs an enterprise innovation perspective to investigate the impact of the reform on micro-enterprise behavior, thereby providing a basis for policy evaluation. This study employs a staggered difference-in-differences (DID) model to examine the influence of the registration system reform on enterprise R&D investment, utilizing both approval-based and registration-based IPO system samples. The findings indicate that, in general, companies listing through the registration-based IPO system do not increase their R&D investment in the year of listing. However, they do increase their R&D investment over time in the three years after listing. This suggests that the registration system reform can effectively promote enterprise innovation, with the effect manifesting as a lag and cumulative effect.

Following an in-depth examination of the registration-based IPO systems in diverse stock markets, the study reveals that the promotional impact of the registration system reform on enterprise innovation is not discernible in the initial year of listing for enterprises on the STAR Market. However, the effect becomes increasingly pronounced and persists over time, becoming more significant in the subsequent three years following listing. In contrast, the registration system reform has had no significant impact on innovation investment in the three years following the listing of enterprises on the ChiNext Market. Furthermore, it significantly inhibits innovation investment in the year of listing. This discrepancy may be attributed to substantial differences in the post-listing continuous supervision methods between the STAR Market and the ChiNext Market. To illustrate, the STAR Market sets more rigorous standards for continuous information disclosure and supervision of sponsor institutions, particularly regarding scientific research investment and other pertinent data. Its investment mechanism is also more stringent, which to some extent standardizes the behavior of listed enterprises, enabling them to maintain robust scientific research capabilities and technical standards. Consequently, in contrast to enterprises listed via the approval-based IPO system, the R&D investment of IPO enterprises on the STAR Market can continue to grow after listing. However, the registration system of the ChiNext Market does not specify clear requirements regarding the disclosure of R&D investment information. Moreover, in comparison to the approval-based IPO system, the delisting system on the ChiNext Market is more rigorous under the registration-based IPO system. Consequently, the registration system reform may prompt IPO enterprises on the ChiNext Market to prioritize profit indicators after listing, which could potentially impede enterprise innovation.

This paper further investigates the mechanisms through which the registration system reform affects enterprise innovation. First, the results demonstrate that the reform of the registration system can lead to an improvement in the turnover of total assets of enterprises, effectively alleviate agency problems between shareholders and managers, and reduce management control over R&D activities, thereby promoting enterprise innovation. Second, compared with SOEs, the impact of registration system reform on R&D investment is more pronounced in non-state-owned enterprises facing heightened financing constraints, suggesting that the reform can facilitate enterprise innovation by alleviating corporate financing constraints. Third, the promotion effect of the registration system reform on R&D investment is more significant for enterprises with a low proportion of technical personnel than for those with a high proportion of technical personnel at the time of listing. This indicates that the registration system reform can help enterprises accumulate innovative human capital and promote enterprise innovation. In conclusion, the above results indicate that reducing agency costs, alleviating financing constraints, and attracting external innovation talents are important channels through which registration system reform promotes enterprise innovation.

This paper contributes to the literature on the relationship between going public and innovation by elaborating on the different impacts of two IPO systems (registration system and approval system) on enterprise innovation in emerging markets. Additionally, this study has some practice implications for regulators, investors and enterprises.

First, it is of paramours important for regulatory authorities and policymakers to gain a comprehensive understanding of the systems in place at each board and to implement differentiated supervision. This is because the response effects of R&D investment intensity vary across different boards under the diversified registration-based IPO system. This may be attributed to the inherent differences in the systems of each board. It is essential to conduct comprehensive assessments of the policy effects by meticulously examining the economic consequences of the registration system reform in each board. Moreover, enhancing publicity, providing comprehensive and detailed policy interpretations, and facilitating investors’ accurate understanding of the extant IPO system, especially the institutional differences across different boards, can boost the implementation of the reform. In addition, this study demonstrates that the registration system reform can promote enterprise innovation by further alleviating the agency problem caused by information asymmetry. Investors have a higher degree of dependence on information disclosure under the registration system [45]. It is therefore recommended that the regulatory authorities reinforce the requirements for the standard degree and quality of information disclosure for IPO companies, strictly enforce the legislation and increase the penalties for disclosure violations. This will protect the legitimate rights and interests of investors, and facilitate the registration system reform to better serve the capital market. Moreover, the significant inhibitory effect of the registration system reform on innovation among ChiNext Market IPO companies may indicate the presence of certain R&D whitewashing behaviors among these companies [9]. It is therefore recommended that relevant departments should implement more rigorous supervision of enterprises following listing on the ChiNext Market. This should include standardizing the use of financing funds, encouraging enterprises to strengthen innovation and enhance core competitiveness, and discouraging the practice of whitewashing behavior prior to listing, to safeguard the rights and interests of investors.

Second, for those considering investment, it is recommended that potential investors gain a comprehensive understanding of the current registration-based IPO system and the associated implications for enterprise innovation in the STAR Market and the ChiNext Market. It is further advised that investors pay particular attention to the differences in listing systems and trading methods across different boards and improve their investment decision-making abilities.

Finally, this paper demonstrates that the registration-based IPO has the potential to markedly enhance the level of R&D investment among listed companies. This is due to the diminished significance of the substantive audit before listing, the increased emphasis on information disclosure within the registration-based IPO system, and the more market-oriented pricing system of the registration system, which places a premium on innovation. Therefore, enterprises must adapt to market changes, capitalize on the opportunities presented by the current era, and prioritize innovation to enhance their core competitiveness. In addition, given the delayed and cumulative promotion effect of the registration-based IPO system on enterprise innovation, listed companies are encouraged to maintain enthusiasm for innovation, leverage the advantages of the capital market to alleviate financing constraints, proactively improve corporate governance, and enrich research and development teams, to promote the sustainable and high-quality development of corporate innovation.

It should be noted that this paper has some limitations. First, given that the registration system has only recently been implemented in China’s capital market, it is not yet possible to test the long-term impact of this reform on corporate innovation. As the implementation period of the registration system increases, future research can further examine the long-term impact of the registration system reform on corporate innovation. Second, when examining the impact of the registration system reform of the STAR Market on corporate innovation, it should be noted that the registration system has been in place since the inception of the STAR Market. Therefore, we can only take the main board as its control group. It may therefore be the case that the differences in R&D investment between listed companies on the STAR Market and on the main board are not a consequence of the differences in the IPO system, but rather reflect the inherent distinctions between the two boards. Once the registration system has been fully implemented in China’s capital market, future research will be able to examine the impact of the registration system reform in the main board market on corporate innovation.