
==== Front
Heliyon
Heliyon
Heliyon
2405-8440
Elsevier

S2405-8440(24)12225-6
10.1016/j.heliyon.2024.e36194
e36194
Research Article
Fund holdings and real earnings management: An empirical analysis of the Chinese A-share market
Lian Lanlan a
Zhao Di b
Dong Qingli itissunny525@gmail.com
b⁎
a School of Information and Business Management, Dalian Neusoft University of Information, Dalian, China
b School of Economics and Management, Dalian University of Technology, Dalian, China
⁎ Corresponding author. School of Economics and Management, Dalian University of Technology, No. 2 Lingshui Road, Dalian, 116025, China. itissunny525@gmail.com
13 8 2024
30 8 2024
13 8 2024
10 16 e3619430 3 2024
9 8 2024
12 8 2024
© 2024 The Authors
2024
https://creativecommons.org/licenses/by-nc/4.0/ This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).
Drawing on data from Chinese A-share listed companies spanning 2016 to 2023, this study investigates how fund holdings influence real earnings management (REM) practices. The results indicate that: (1) There is a significant negative correlation between fund holdings and REM, suggesting strong external oversight; (2) The governance effect of fund holdings is more pronounced in non-state-owned firms and those with a balanced equity structure, underscoring the critical roles of ownership nature and equity distribution; (3) High-quality internal controls also exhibit a significant negative correlation with REM, serving as a mediating factor between fund holdings and earnings management. These findings deepen our understanding of how institutional investors contribute to corporate governance and enhance internal control systems.

Keywords

Fund holdings
Real earnings management
Internal control quality
Corporate governance
==== Body
pmc1 Introduction

Earnings management is a significant technique employed by listed companies to adjust profits and embellish performance, and it has been a persistent challenge in capital markets [1]. The root of earnings management lies in the “communication blockages” and “contractual frictions” arising from principal-agent problems [2]. Since accounting information from listed companies serves as a cornerstone for effective resource distribution in capital markets [3], any manipulation of earnings data distorts the true economic performance of companies, leading to significant disruptions in market resource allocation. The opportunistic behavior of earnings management not only harms the interests of investors at an informational disadvantage but also adversely affects the long-term value of companies. The supervision of corporate earnings management primarily involves two aspects: external governance and internal governance [4]. A strong framework of both external and internal governance strengthens oversight within a company, limiting the incentives and abilities of management and major shareholders to engage in earnings management, which in turn enhances the quality of accounting information. External governance includes legal regulations, standards, and independent audits, while internal governance factors refer to internal mechanisms related to comprehensive corporate governance. Currently, as the external governance environment in China's capital markets becomes increasingly stringent, managers might seek more opportunities within the gaps of internal governance. Therefore, studying the role of internal governance factors in the operational space for earnings management becomes increasingly important.

Since the early 21st century, when the China Securities Regulatory Commission (CSRC) introduced the “Strategic Development of Institutional Investors,” the Chinese capital market has witnessed rapid growth in institutional investors [5]. Fig. 1 depicts the proportions and evolving trends of institutional versus individual investors in the A-share market over the last three years, highlighting the dominant position and influence of institutional investors in the Chinese capital market. Institutional investors, with their substantial informational, financial, and expertise-related advantages over individual investors, are better equipped and more motivated to oversee the actions of management and controlling shareholders in listed companies. This oversight helps to curb opportunistic behaviors, improve corporate governance effectiveness, boost company value, and ultimately maximize investment returns [6]. The main types of institutional investors encompass securities investment funds (both public and private), securities firms, insurance companies, Qualified Foreign Institutional Investors (QFII), social security funds, banks, and various other institutions. According to Wind1 data, as of the end of 2023, fund holdings accounted for the highest proportion among institutional investors, reaching 11.49 %, as shown in Fig. 2. In comparison to other categories of institutional investors, the “core business” of fund holdings is selecting and holding stocks of listed companies, making fund holdings particularly prominent in terms of specialization. Although direct impacts of fund holdings on corporate earnings management are rarely observed in practice, there are numerous real cases reflecting the substantial involvement of funds in corporate governance,2,3 This raises an intriguing question about whether fund holdings can influence corporate earnings management.Fig. 1 Composition of investors in China's capital market over the last three years.

Fig. 1

Fig. 2 Breakdown of institutional investors in China's capital market for 2023.

Fig. 2

Current research on the impact of institutional investors on earnings management governance remains inconclusive, with three main competing perspectives emerging [7]. The first viewpoint posits that institutional investors do not significantly enhance corporate governance and may even exacerbate management's earnings management behavior [8,9]. The second viewpoint supports the supervisory role of institutional investors, arguing that they can effectively constrain management's opportunistic behaviors [10]. The third perspective posits that the governance effectiveness of institutional investors is largely dependent on their shareholding proportions [11]. Additionally, treating institutional investors as a homogeneous group in research overlooks their heterogeneity, which is crucial for accurately reflecting the internal governance effects of institutional investors [12]. Research shows that institutional investors are adopting increasingly diverse investment styles, resulting in notable differences in their corporate governance behaviors. Consequently, the role of institutional investors as “effective supervisors” is not consistently observed in practice [13,14]. Conversely, REM has emerged as a major challenge for enterprises. Since the adoption of International Financial Reporting Standards (IFRS) by Chinese companies in 2006, shifts in accounting standards have prompted a move from accrual-based earnings management to real activities manipulation, a practice that is harder to regulate. This shift has significantly increased the concealment of earnings management behaviors [15,16]. These practices not only skew business decisions but also damage the interests of shareholders and other stakeholders, underscoring the critical need to examine the governance impact of internal controls on REM.

Building on this background, the study seeks to explore the following research questions: 1). How do fund holdings, as a key element of institutional investors, contribute to corporate governance? 2). Does fund ownership have a governance effect on REM in enterprises? 3). What is the transmission path of this governance effect?

To explore these research questions, this study analyzes data from Chinese A-share listed companies spanning 2016 to 2023, conducting an empirical investigation into the governance impact of fund holdings on corporate earnings management. Additionally, the study examines how ownership structure influences the governance impact of fund holdings and delves into the mediating role of internal control quality in this relationship. The findings show that fund holdings lead to a significant decrease in REM, with this effect being especially strong in firms with robust internal controls, notably in non-state-owned enterprises and companies where the largest shareholder lacks absolute control. This study aims to deepen the understanding of the role of fund holdings in corporate governance within the Chinese capital market, while also providing policy recommendations for regulatory authorities and investors. This study not only adds depth to the literature on the connection between institutional investors and REM but also sheds light on the governance role of fund holdings and broadens the understanding of internal control quality's impact.

The marginal contributions of this study are highlighted in the following areas: First, unlike previous studies that view institutional investors as a homogeneous group, this research specifically examines fund holdings, analyzing their supervisory impact and offering new insights into the role and value of institutional investors. This perspective helps reveal the specific roles and impacts of fund companies in corporate governance. Second, this research deepens the understanding of the governance impact of fund holdings through the lens of earnings management. Unlike earlier studies that concentrate on the returns and risks of securities investment funds, this research explores the role of fund holdings in mitigating earnings management behaviors, offering a thorough analysis that includes ownership structure and internal control quality. This provides new insights into how fund holdings influence corporate behavior through internal governance mechanisms. Lastly, this study offers empirical evidence and policy recommendations for practitioners, particularly regulatory authorities, in optimizing the governance structure of institutional investors, enhancing the transparency of listed companies, and improving internal control quality.

The structure of the paper is as follows: The second section reviews the relevant literature and introduces the research hypotheses. The third section outlines the research design, covering sample selection, data sources, variable definitions, and model construction. The fourth section presents the empirical results and discussion, while the final section summarizes the key findings and offers policy recommendations based on the study.

2 Literature review and research hypotheses

2.1 Governance effect of fund holding on earnings management

The division between ownership and control in contemporary enterprises creates agency issues between management and shareholders [17]. Motivated by various factors such as compensation contracts, debt contracts, and political incentives, the management has the motive to engage in opportunistic behaviors that harm the company's value [18]. These agency conflicts lead to contract frictions, which are the fundamental drivers of earnings management. Information asymmetry exacerbates this phenomenon, making it difficult to eliminate earnings management [19]. Rational investors will inevitably seek effective governance mechanisms to restrain management. However, investors who carry out supervision bear all the costs while supervision benefits are shared by all investors. This mismatch between costs and benefits leads to a widespread “free-rider” problem.

Institutional investors, as special participants in the capital market, have three significant characteristics: first, institutional investors have a positive supervisory motivation. Due to their large shareholding proportion, the benefits generated from supervision can offset the cost of supervision and effectively alleviate the problem of small shareholders “free-riding” under dispersed equity ownership. Second, institutional investors have stronger supervisory capabilities. They possess more talent, resources, and investment experience; they also access diverse channels for information acquisition—including financial statements, market research, industry studies— and typically engage directly with company management and directors. Third, institutional investors are more rational compared to individual investors. Their higher shareholding ratio enables them to “vote with their feet” when management performs poorly, thereby increasing the likelihood of stock price decline. Listed companies seek stability in stock prices and hope for long-term support from institutional investors; therefore, institutional investors wield strong influence over management.

Due to significant differences in the scale of funds, development level, and governance motivation among different types of institutional investors, conclusions based solely on a general examination of institutional investors may be questioned [20,21]. In recent years, there has been increasing research focus on the diversity among institutional investors. Most existing studies categorize institutional investors based on distinctions such as long-term versus short-term, active versus passive, and trading-oriented versus stable-oriented [12,22,23], and some research suggests that long-term institutional investors tend to curb earnings management, whereas short-term investors often push management to prioritize short-term profits, leading to increased earnings management [24]. Overall, various types of institutional investors have distinct stock holding preferences and assume different roles in corporate governance. Considering the diversified development of institutional investors, we believe that they should not be simply labeled. As shown in Fig. 2, fund holdings are an important part of China's capital market and hold a pivotal position among institutional investor groups. Therefore, this study focuses on the governance effects of fund holdings from a micro perspective which holds significance for China's capital market's green transformation as well as guiding role for securities investment funds.

Existing research indicates that fund holdings influence corporate governance in two key areas [25]. On one hand, fund holdings restricts the behavior of large shareholders encroaching on the rights of minority shareholders; on the other hand, fund holdings supervises management behavior and decision-making [26]. The supervision of management corresponds to the agency problem between shareholders and management. Management is responsible for operational management of the company, with a significant information advantage that can easily manipulate financial information through earnings management to seek personal gain. Funds are inherently profit-driven and may strengthen their oversight of management to constrain their earnings management behaviors because participating in governance brings greater benefits than “voting with their feet”. Funds pool retail capital, have greater influence, and have a stronger incentive to participate in supervising management to obtain monitoring benefits. They provide an independent buyer force as agents for minority shareholders, effectively counterbalancing large shareholders and restraining their earnings management behaviors which infringe upon minority shareholder interests [27]. Therefore, securities investment funds possess both motivation and capability to participate in managing earnings within listed companies.

Given this context, we anticipate that fund holdings will effectively oversee the management of listed companies, particularly in curbing REM practices. Accordingly, we propose the first hypothesis.H1 Fund holdings can restrain a company's REM behavior.

2.2 The influence of equity structure on the governance effect of fund holdings

Equity structure determines the distribution of power and allocation of benefits within a company. Within the framework of the Chinese institutional environment, this study examines how ownership structure impacts the governance effect of fund holdings, focusing on two aspects: ownership balance and ownership nature.

2.2.1 Analysis from the perspective of equity balance

The equity structure determines the way and effect that shareholders exercise power, thereby affecting governance effectiveness. For example, Saleem Salem Alzoubi found that controlling shareholders may disclose surplus information according to their preferences and interests, thereby weakening information content [28]. Similar conclusions were drawn in recent studies [29,30]. Masulis et al. argue that large shareholders might pursue excessive control rights to gain private benefits, with this expropriation drive being more pronounced when there is a significant separation between ownership and control [31]. If multiple large shareholders coexist and check each other, it will lead to a dispersed shareholding ratio, reduce supervision costs effectively, and mitigate the second type of agency problem. Margaritis and Psillaki argue that having multiple major shareholders with relatively even holdings can help enhance company value [32]. Andrias suggests that equity balance can have positive effects in terms of property rights – if several controllers jointly hold higher cash flow ownership over assets belonging to companies they control which would increase transfer costs for them reducing private returns on their control rights ultimately increasing company value [33]. Boubaker and Sami's study also demonstrates that the presence of multiple major shareholders significantly improves the information content of earnings [34]. It is thus evident that equity balance has received considerable support for its positive influence on corporate governance effectiveness.

The impact of fund holdings on earnings management may be affected by the level of equity balance. In China, the equity structure of listed companies is typically highly concentrated, with controlling shareholders frequently maintaining significant control over the company. The control advantages enjoyed by controlling shareholders can result in opportunistic tunneling behaviors that negatively impact the interests of minority shareholders. However, when multiple controlling shareholders coexist and establish a cooperative supervisory framework, they are motivated and capable of overseeing management, thereby reducing the first type of agency problem and simultaneously safeguarding minority shareholders’ interests, which helps to alleviate the second type of agency problem. Therefore, equity balance is considered an effective governance mechanism. In companies with a dominant shareholder, the proportion of fund holdings relative to the largest shareholder is typically smaller, thus weakening their influence. Conversely, in companies with a higher degree of equity balance, securities investment funds generally have greater voice and influence. Based on this, we propose that the governance effect of fund holdings will be affected by the degree of equity balance. Accordingly, we put forward the following hypothesis.H2 The degree of equity balance affects the governance effect of fund holdings on earnings management. A highly balanced equity structure enhances the governance effect of fund holdings.

2.2.2 Analysis from the perspective of equity nature

The characteristics of equity holders, particularly the identity of the largest shareholder, have a substantial influence on the governance impact of fund holdings. In the unique institutional environment of China, it is essential to consider the influence of corporate nature when discussing issues related to corporate governance and capital markets. Current research offers two opposing perspectives on how equity nature affects corporate governance. Some studies argue that state-owned enterprises (SOEs) prefer to gain local government support, resulting in positive governance outcomes [35]. Conversely, other studies argue that SOEs are more prone to ownership absence issues, resulting in relatively high agency costs and limited effectiveness of conventional internal and external governance mechanisms [36]. Typically, SOEs exhibit more concentrated equity structures than non-SOEs, which often results in higher levels of REM. According to the analysis in the previous section, this concentration weakens the voice of fund holders and limits the role of funds in corporate earnings management. Moreover, the presence of state shareholders in SOEs, often represented by entities like the State-owned Assets Supervision and Administration Commission (SASAC), leads to issues of ownership absence and insider control, which further weakens the internal oversight of REM by SOE shareholders.

Given the significant differences in incentive motivations and functional responsibilities between SOEs and non-SOEs, we believe that the governance effect of fund holdings on earnings management will be influenced by the identity of equity holders. Firstly, owners of non-SOEs enjoy full residual claims, providing stronger incentives to enhance corporate value. In contrast, SOEs prioritize not only economic value but also social responsibilities and political objectives, which heavily influence their management's economic decisions. Secondly, SOEs bear responsibilities such as safeguarding national and social needs, regulating the economy, and promoting employment. With long-term access to government resources and protection, SOEs are less reliant on the capital market, reducing management's sensitivity to external supervision mechanisms like fund holdings. As a result, we hypothesize that the governance impact of fund holdings is more pronounced in non-SOEs, leading to the following hypothesis.H3 The identity of equity holders influences the governance impact of fund holdings on earnings management. Fund holdings have a more significant governance effect in non-SOEs.

2.3 Mechanism effect analysis

Ensuring the accuracy and reliability of accounting information is a crucial objective of internal control. Robust internal controls can significantly reduce earnings management behaviors, address agency issues stemming from “information asymmetry” and reflect management's dedication to maintaining high-quality financial reporting [37]. Since the introduction of the Sarbanes-Oxley (SOX) Act in the United States in 2002, there has been an increasing global focus on developing and enhancing internal control systems. In China, the continuous rollout and enforcement of the “Basic Norms for Enterprise Internal Control” along with its related guidelines have greatly enhanced the internal control framework. Research indicates that a robust internal control system not only ensures the authenticity and accuracy of financial information but also significantly reduces the occurrence and extent of REM [38]. For instance, research utilizing internal control attestation reports as an indicator of internal control quality has shown that strong internal controls can significantly reduce earnings management within companies [39]. Moreover, the presence of internal control deficiencies is negatively correlated with earnings management levels, suggesting that companies with stronger internal controls generally engage in less earnings management [38,40,41].

Experienced institutional investors can effectively constrain and supervise management, prompting them to establish a robust internal control system, thereby reducing earnings management behaviors [42,43]. Specifically, fund companies, as proactive participants in corporate governance, leverage their professional teams and extensive investment experience to directly supervise management and curb earnings management behaviors. They also encourage firms to enhance their internal control quality and improve corporate governance structures. Through this dual role, fund holdings not only directly reduce the motives for managerial earnings management but also indirectly lower the occurrence and extent of REM by improving internal control quality. This study therefore suggests that fund holdings serve as external supervisors, contributing to corporate governance by curbing earnings management. Additionally, they enhance the quality of internal controls, which helps to alleviate information asymmetry within the company and further reduce earnings management. In other words, fund holdings improve the quality of internal controls, which in turn provides a mediating mechanism through which fund holdings influence earnings management. Building on this premise, we propose the following hypotheses.H4a The higher the quality of internal control, the lower the level of REM in enterprises.

H4b Internal control quality mediates the effect of fund holdings on suppressing earnings management in enterprises.

3 Research design

3.1 Sample selection and data sources

This study utilizes a sample of Chinese A-share listed companies from 2016 to 2023. To ensure the accuracy and stability of the results of this study, the following filters and treatments were applied to the collected raw data.(1) Exclude observation samples of listed financial enterprises;

(2) Exclude observation data from ST and *ST listed companies;

(3) Exclude observation samples of listed companies that have not disclosed financial data or have abnormal values;

(4) To prevent inaccuracies in research results caused by outlier data within the large dataset, this study employs the Winsorization method to tail-trim all continuous data (at the 1 % and 99 % percentiles).

Table 1 displays how we selected our sample. Initially, we collected data on 3458 publicly listed companies in China A shares. We excluded 90 financial institutions and an additional 227 companies that were designated ST and *ST by the China Securities Regulatory Commission (CSRC) during the sample period. Furthermore, due to varying degrees of missing data across different variables, we eliminated another 1450 companies for data disclosure issues. The final sample comprised 1691 companies, yielding a total of 13,523 firm-year observations for empirical analysis.Table 1 Sample selection procedure.

Table 1Sample selection	No.	
Initial Chinese A-share listed firms from 2016 to 2023	3458	
Less:	
 Financial firms	90	
 Firms labeled as ST/*ST as of the end of the fiscal years	227	
 Firms that have not disclosed financial data or have abnormal valuesa	1450	
Firms in the final sample	1691	
Firm-year observations in the final sampleb	13523	
a We removed samples of enterprises with a higher amount of missing data due to inconsistencies and variable distribution.

b During the final process of sorting panel data, we discovered five abnormal observations for firm-year and subsequently removed them.

Empirical analyses of the data were conducted using Stata 16.0 and SPSS 26.0; the internal control index data were derived from the DIB database,4 while other data variables were obtained from the CSMAR database.5

3.2 Variables

3.2.1 Dependent variable

The dependent variable in this study is REM, which existing research typically measures using three sub-indicators: abnormal operating cash flows (UCFO), abnormal discretionary expenses (UDISX), and abnormal production costs (UPROD) [44,44,45]. However, different scholars have different measurement methods.

Firstly, normal operating cash flows (CFO), normal discretionary expenses (DISX), and normal production costs (PROD) are estimated through annual and industry-specific regression models, with the specific models as follows:(1) CFOitTAi,t−1=α1+α21TAi,t−1+α3SALESitTAi,t−1+α4ΔSALESitTAi,t−1+εit

UCFO is then calculated as the difference between the actual CFO and the normal level of CFO derived from the estimated coefficients in Eq (1).

In theory, discretionary expenses should be defined as a function of current period sales revenue and the previous period's total assets. However, modeling discretionary spending based on current sales presents a mechanical issue. If a firm manipulates its sales to boost earnings in a certain year, it will significantly decrease the model residuals [44]. To resolve this issue, we model discretionary expenses using lagged sales and employ the following model to estimate a “normal” level of discretionary expenses:(2) DISXitTAi,t−1=α1+α21TAi,t−1+α3SALESi,t−1TAi,t−1+εit

UDISX is then determined by subtracting the normal level of DISX, calculated using the estimated coefficients from Eq (2), from the actual DISX.

Production costs are defined as the total of the cost of goods sold (COGS) and changes in inventory over the year. COGS can be modeled as a linear function of contemporaneous sales:(3) COGSitTAi,t−1=α1+α21TAi,t−1+α3SALESitTAi,t−1+εit

Next, inventory growth can be modeled as a linear function of both current and lagged changes in sales as follows:(4) ΔINVitTAi,t−1=α1+α21TAi,t−1+α3ΔSALESitTAi,t−1+α4ΔSALESi,t−1TAi,t−1+εit

Based on Eq (3) and Eq (4), we can estimate the normal production costs (RPROD) as:(5) PRODitTAi,t−1=α1+α21TAi,t−1+α3SALESitTAi,t−1+α4ΔSALESitTAi,t−1+α5ΔSALESi,t−1TAi,t−1+εit

UPROD is then calculated by subtracting the normal level of PROD, determined using the estimated coefficients from Eq (5), from the actual PROD.

The variables involved in the above equations have been summarized in Table 2. Among them, CFO is the operating cash flow of firm i during period t; DISX refers to the discretionary expenditures of firm i during period t, including advertising expenses, R&D expenses, and SG&A. PROD represents the production costs of firm i during period t, comprising the sum of COGS and inventory changes. UCFO, UDISX, and UPROD are determined by calculating the difference between the actual values and the normal levels predicted by Eqs (1), (2), (5). These three variables are used as proxies for REM. Under the same sales level, upward earnings management may have one or more of the following situations: unusually low operating cash flow, and/or unusually low discretionary expenditures, and/or unusually high production costs.Table 2 Definitions and variables of the REM model.

Table 2Variable	Definition	Source	
CFOit	Cash flow from operating activities of firm i during period t	CSMAR	
TAi,t−1	Total assets of firm i at the end of period t−1	CSMAR	
SALESit	Sales revenue of firm i during period t	CSMAR	
ΔSALESit	Change in sales revenue of firm i from period t−1 to period t	CSMAR	
ΔSALESi,t−1	Change in sales revenue of firm i from period t−2 to period t−1	CSMAR	
PRODit	Production costs of firm i during period t, which is the sum of the current period's COGS and changes in inventory	CSMAR	
DISXit	Discretionary expenses of firm i during period t, defined as the sum of advertising expenses, R&D expenses and SG&A	CSMAR	

To capture the overall impact of REM, we integrate three individual measures to comprehensively assess REM activities. Unlike [45], we first adopt a mathematically intuitive measurement method: REM = UPROD - UCFO - UDISX. This method implies that upward earnings management may involve one or more of the following situations: unusually low operating cash flow, discretionary expenditures, or high production costs. The REM measured by this method fully considers the monotonic differences of its different components. A higher value in this aggregate measure suggests a greater probability that the firm is involved in REM activities. Naturally, we also took into account the conventional REM measurement methods found in previous literatures [[44], [45], [46]]. These include REM2 = - UCFO - UDISX and REM3 = UPROD + UCFO + UDISX. We then tested these methods during the subsequent robustness analysis phase.

We recognize that although REM can be computed using the aforementioned measures, the three individual variables that underlie it may have varying effects on earnings. This could potentially dilute any results using the aggregated measures, regardless of which one is employed. Consequently, we provide results for both the aggregate REM measure and the three individual proxies for REM (UCFO, UPROD, and UDISX).

3.2.2 Explanatory variables, mediating variables, and control variables

The explanatory variable is the proportion of fund holdings, while the mediating variable is the quality of internal control. These are calculated as follows:

Fund Holdings (Fund): Measured as “end-of-year fund holdings/total number of shares at year-end * 100 %” to determine the fund holdings ratio (Fund).

Internal Control Quality (Dib): Building on existing research, this study assesses the quality of a company's internal controls by using the natural logarithm of the internal control index obtained from the DIB database [47].

To ensure the stability and reliability of the conclusions of this study, following existing research [16,43,[48], [49], [50]], this study introduces leverage (Lev), return on equity (Roe), company growth (Growth), ownership concentration (Top1), audit firm (Big4), company size (Size), time dummy variables (Year), and industry dummy variables (Industry) as control variables.

Table 3 outlines the types, symbols, and definitions of all variables utilized in this study.Table 3 Variable definition.

Table 3Variable	Symbol	Description	Source	
Real Earnings Management	REM	Abnormal production costs - Abnormal cash flows from operating activities - Abnormal expenses	CSMAR	
Fund Holdings Ratio	Fund	Fund holdings/Total number of shares	CSMAR	
Internal Control Quality	Dib	Natural logarithm of the internal control index from Dibo Company	DIB	
Leverage Ratio	Lev	Total liabilities at the end of the period/Total assets at the end of the period	CSMAR	
Return on Equity	Roe	Net profit/Net assets	CSMAR	
Company Growth	Growth	(Current year's operating income - Last year's operating income)/Last year's operating income	CSMAR	
Ownership Concentration	Top1	Shareholding percentage of the largest shareholder	CSMAR	
Audit Firm	Big4	Big4 = 1, indicating the company is audited by one of the Big Four accounting firms; Big4 = 0, indicating the company is not audited by one of the Big Four accounting firms	CSMAR	
Company Size	Size	Natural logarithm of total assets	CSMAR	
Time	Year	Dummy variable	CSMAR	
Industry	Ind	Dummy variable	CSMAR	

3.3 Model construction

To examine the effect of fund holdings on REM, the study uses the ratio of fund holdings as the explanatory variable and REM as the dependent variable. To test hypothesis H1, the following baseline regression model is constructed as Eq (6):(6) REMit=α1+α2Fundit+α3CONTROLSit+πi+λt+εi,t

where i and t respectively represent the firm and the year, REMit stands for the level of REM of the firm, Fundit represents the level of fund holdings of the firm, CONTROLSit refers to a series of control variables. πi and λt denote industry fixed-effects and year-fixed effects, respectively. εi,t is the random error term. We chose the fixed effect model because it can handle panel data effectively, control unobservable individual effects, eliminate omitted variable bias, manage heterogeneity among individuals and improve causal inference capabilities. This enhances the accuracy and reliability of estimation results. In this study, the fixed effect model was employed for both the benchmark regression analysis and for robustness and heterogeneity analyses.

To further investigate the relationships between the fund holdings ratio and each of the components of REM, namely abnormal operating cash flows, abnormal expenses, and abnormal production costs, this study builds the following sub-models, as Eqs (7), (8), (9), based on the baseline model:(7) UCFOit=α1+α2Fundit+α3CONTROLSit+πi+λt+εi,t

(8) UDISXit=α1+α2Fundit+α3CONTROLSit+πi+λt+εi,t

(9) UPRODit=α1+α2Fundit+α3CONTROLSit+πi+λt+εi,t

To test hypothesis H2, this study categorizes the samples into two groups: “Absolute control by major shareholders” and “Non-Absolute control by major shareholders”. The criterion for classification is based on whether the largest shareholder's shareholding exceeds 50 %, which reflects differences in equity checks and balances. By comparing α2 values across these two groups, we can determine if hypothesis H2 holds true.

Besides, we conduct group tests to examine the impact of equity on governance in SOEs and non-SOEs. Through observing α2 values across different sample groups, we can test the establishment of hypothesis H3.

To validate hypothesis H4a, this study constructs the following model, as Eq (10), concerning the quality of internal control and REM:(10) REMit=α1+α2Dibit+α3CONTROLSit+πi+λt+εi,t

Finally, to validate Hypothesis H4b, which examines the mediating role of internal control quality in the baseline regression, this study constructs the following models, Eqs (11), (12):(11) Dibit=α1+α2Fundit+α3CONTROLSit+πi+λt+εi,t

(12) REMit=α1+α2Fundit+α3Dibit+α4CONTROLSit+πi+λt+εit

4 Empirical results and discussion

4.1 Descriptive statistics

Table 4 presents the descriptive statistics for the variables studied. Observing the range of REM, the minimum value is −2.098 and the maximum value is 1.959. This wide distribution range indicates that different forms of REM behavior are common among listed companies. Specifically, for each indicator, the average values of abnormal operating cash flows (UCFO), abnormal expenses (UDISX), and abnormal production costs (UPROD) are 0.0005, 0.0003, and −0.0007 respectively, further confirming the widespread phenomenon of REM among listed companies. For Fund, the average is 3.196 % and the median is 1.417 %, which indicates that Fund typically hold a small proportion of shares. However, there is significant variation in these holdings, with the fund holding ratio ranging from as low as 0.0001 % to as high as 38.042 %. We also calculated the Interquartile Range (IQR) of the Fund to show the range between its 75th percentile (Q3) and its 25th percentile (Q1), indicating the middle 50 % of Fund. This measure provides insight into data dispersion and is not influenced by outliers. These findings all point to a polarization in market distribution of Fund. The average internal control quality (Dib) is 6.237, with a maximum of 6.847, suggesting that most listed companies have a medium level of internal control quality, leaving room for improvement. Regarding auditing, the data indicates that, on average, 6 % of the sample companies are audited by Big Four accounting firms (Big4), implying that the majority are audited by non-Big Four firms. Additionally, the average value for ownership concentration (Top1) is 33.003 %, with the maximum reaching 89.090 %, indicating a prevalent phenomenon of high ownership concentration among Chinese listed companies, that is, a “single dominant shareholder”. The Kolmogorov-Smirnov (KS) test results for each indicator in Table 4 indicate that all variables rejected the null hypothesis of normal distribution at the 1 % level, indicating that they do not follow a normal distribution.Table 4 Descriptive statistics for the main variables.

Table 4Variable	No. of obs	Min	Max	Mean	Median	IQR	Std.	KS	
REM	13523	−2.098	1.959	−0.001	0.01	0.178	0.186	0.081***	
UCFO	13523	−0.423	0.426	0.0005	0.001	0.074	0.070	0.056***	
UPROD	13523	−1.593	1.727	−0.0007	0.004	0.097	0.112	0.094***	
UDISX	13523	−0.364	0.412	0.0003	−0.007	0.044	0.060	0.151***	
Fund	13523	0.0001	38.042	3.196	1.417	4.08	4.432	0.236***	
Dib	13523	0	6.847	6.237	6.488	0.126	1.194	0.42***	
Lev	13523	0.008	3.919	0.434	0.426	0.3	0.205	0.033***	
Size	13523	17.779	28.636	22.458	22.281	1.627	1.323	0.061***	
Growth	13523	−1.309	429.036	0.300	0.1	0.284	4.323	0.409***	
Roe	13523	−186.557	71.765	−0.006	0.064	0.083	2.602	0.439***	
Top1	13523	2.870	89.090	33.003	30.63	20.5	14.497	0.066***	
Big4	13523	0	1	0.060	0	0	0.237	0.54***	

To confirm the suitability of the chosen control variables, this study conducted a collinearity diagnosis. The results indicate that the variance inflation factor (VIF) for all control variables is below 2, signaling no issues with collinearity. Detailed findings can be found in the last column of Table 5. Furthermore, Table 5 displays the Spearman correlation coefficients between the variables. The decision not to use the Pearson correlation coefficient stems from the prior KS test results demonstrating that these variables do not adhere to a normal distribution. The correlation coefficient between fund holdings and REM is −0.17, significant at the 1 % level, providing initial support for hypothesis H1 that there is a negative correlation between fund holdings and REM behavior.Table 5 Spearman correlation coefficients of the main variables.

Table 5	REM	UCFO	UPROD	UDISX	Fund	Dib	Size	Lev	Roe	Growth	Top1	Big4	VIF	
REM	1	−0.661***	0.885***	−0.594***	−0.17***	−0.119***	0.189***	0.019**	−0.118***	−0.348***	−0.052***	−0.058***		
UCFO		1	−0.396***	0.092***	0.135***	0.104***	−0.164***	0.011	0.044***	0.317***	0.074***	0.057***	1.217	
UPROD			1	−0.499***	−0.157***	−0.102***	0.201***	0.049***	−0.097***	−0.344***	−0.036***	−0.036***	1.605	
UDISX				1	0.095***	0.064***	−0.017**	0.013	0.167***	0.097***	0.003	0.053***	1.392	
Fund					1	0.231***	0.016*	0.305***	0.198***	0.314***	−0.023***	0.075***	1.087	
Dib						1	−0.019**	0.182***	0.354***	0.448***	0.15***	0.127***	1.053	
Size							1	0.505***	0.002	−0.07***	0.053***	0.118***	1.423	
Lev								1	0.041***	0.147***	0.167***	0.269***	1.6	
Roe									1	0.332***	−0.003	−0.005	1.033	
Growth										1	0.182***	0.085***	1.012	
Top1											1	0.131***	1.086	
Big4												1	1.155	
Note: *, **, and *** indicate that the correlation coefficient is significant at the 10 %, 5 %, and 1 % levels, respectively. Below is the same.

Furthermore, the correlation between fund holdings and abnormal operating cash flows, abnormal expenses, and abnormal production costs has also been confirmed. The relationship between internal control quality and REM is negatively correlated, with significant correlations also observed between internal control quality and abnormal operating cash flows, expenses, and production costs. This indicates that strong internal controls can effectively suppress REM across various dimensions. Additionally, fund holdings are positively correlated with internal control quality and are statistically significant, indicating that as the proportion of fund holdings increases, the quality of the company's internal controls improves. This offers initial evidence supporting the mediating role of internal control quality in the relationship between fund holdings and REM. In the analysis of control variables, the leverage ratio (Lev), company size (Size), company growth (Growth), return on equity (Roe), the presence of the Big Four accounting firms (Big4) are positively correlated with REM, while ownership concentration (Top1) is negatively correlated with REM, further detailing the factors influencing REM behavior.

4.2 Baseline regression analysis

Table 6 displays the regression analysis results on the impact of fund holdings on REM. The findings indicate that the regression coefficient of fund holdings is −0.007, showing a significant negative correlation at the 1 % level. This suggests that a higher proportion of fund holdings is associated with lower levels of REM in listed companies, thereby confirming Hypothesis H1 of this study. This conclusion is in line with existing researches [[25], [26], [27]].Table 6 Regression results of baseline model.

Table 6	REM	UCFO	UPROD	UDISX	
Fund	−0.007*** (−10.41)	0.002*** (10.31)	−0.004*** (−8.98)	0.002*** (7.91)	
Size	0.001 (0.06)	0.003*** (5.92)	0.002** (2.39)	−0.001 (−2.64)	
Lev	0.166*** (10.07)	−0.07*** (−18.98)	0.09*** (10.27)	−0.008*** (−2.7)	
Roe	−0.0001* (−1.723)	0.0004** (2.15)	0.0002*** (0.58)	−0.0008 (−0.44)	
Growth	−0.0008** (−2.1)	−0.0003** (−2.36)	0.0008*** (3.48)	0.002*** (11.4)	
Top1	−0.0008*** (−6.97)	0.0004*** (8.43)	−0.0004*** (−5.34)	0.00006* (1.64)	
Big4	−0.05*** (−7.11)	0.01*** (4.42)	−0.02*** (−5.62)	0.014*** (6.16)	
Hausman	245.58***	214.38***	134.16***	234.33***	
Industry and Year FE	Yes	Yes	Yes	Yes	
No. of obs	13523	13523	13523	13523	
R squared	0.274	0.168	0.153	0.064	

Furthermore, the data in Table 6 also break down the impact of fund holdings on different earnings management behaviors. Specifically, the regression coefficient between fund holdings and abnormal operating cash flows (UCFO) is 0.002, with abnormal production costs (UPROD) at −0.004, and with abnormal expenses (UDISX) at 0.002, all significant at the 1 % level. These results indicate that fund holdings have a comprehensive effect on mitigating the company's REM behavior, especially in controlling abnormal operating activities, costs, and product costs. It is important to note that the coefficient signs for UCFO, UPROD, and UDISX align with our analytical logic in section 3.2.1. At the same level of sales, upward earnings management often involves abnormally low operating cash flow, and/or abnormally low discretionary expenditures, and/or abnormally high production costs. In other words, fund holdings in Table 5 results will restrain earnings management within enterprises, reflected in the inhibitory effect on UPROD and the enhancing effect on UCFO and UDISX. This result logically and evidentially supports using REM = UPROD-UCFO-UDISX to measure earnings management behavior in this study.

Regarding control variables, the analysis reveals some interesting findings: the regression coefficient for leverage ratio (Lev) is 0.166 and positively correlated at the 1 % significance level, suggesting that companies with a higher leverage ratio are more inclined to engage in REM; the regression coefficient for return on equity (Roe) is −0.0001, with a significance level of 10 %, indicating that a higher return on equity may limit management's REM behavior; the regression coefficient for company growth (Growth) is −0.0008, with a significance level of 5 %, implying that companies with stronger growth are less involved in REM; the regression coefficient for the audit firm (Big4) is −0.05, with a significance level of 1 %, reflecting that audits by the Big Four accounting firms can effectively constrain a company's REM behavior.

In summary, fund holdings have a significant negative impact on the REM of listed companies, while good financial conditions and high-quality audit services can further limit REM behavior, thereby enhancing the financial transparency and credibility of the company.

4.3 Analysis of equity structure heterogeneity

Previous studies have shown that REM behaviors not only vary among different institutional investors but also exhibit significant differences across different companies [48,51,52]. In light of this, the study further investigates the heterogeneity, focusing particularly on two aspects: equity balance (i.e., presence of absolute control) and equity nature (i.e., whether the company is a state-owned enterprise).

4.3.1 Equity balance perspective

In Chinese listed companies, shareholding concentration is prevalent, and this ownership concentration generally results in a positive correlation between the proportion of shares held by the largest shareholders and the company's REM practices [28]. Specifically, when shareholding is highly concentrated, major shareholders can have more discretionary power and may engage in earnings management behaviors to seek personal gains, thereby harming the interests of minority shareholders [30,31,33]. Moreover, studies have shown that in Chinese firms, the largest shareholders sometimes utilize real transactions with related parties for earnings management [34,53,54], leading companies with higher major shareholder ownership ratios to be more likely involved in higher levels of REM [51].

However, as analyzed earlier, while funds might participate in corporate governance to suppress REM through supervising management and balancing equity, this suppressive effect could be limited in cases where major shareholders have absolute control. Therefore, this study predicts that in publicly listed companies with high shareholder concentration, the role of funds in curbing REM may not achieve the expected effect.

To achieve this, the study splits the sample into two groups: “absolute control by major shareholders group” and “non-absolute control by major shareholders group”, defining “absolute control by major shareholders group” as when the largest shareholder's holding exceeds 50 %. Before conducting regression analysis, a comparative analysis of the sub-samples was performed. Table 7 highlights significant differences in both the mean and median levels of REM between the two groups. Furthermore, most variables exhibit noteworthy differences in means and medians across sub-samples. These results indicate that our categorization of the sample based on equity balance is statistically significant.Table 7 Sample statistical analysis from the perspective of equity balance.

Table 7	Absolute control by major shareholders	Non-Absolute control by major shareholders	Difference in	
	Mean	Median	Mean	Median	Means (t-stat)	Medians (z-stat)	
REM	−0.023	−0.007	0.002	0.013	−0.025*** (5.292)	−0.02*** (−6.294)	
UCFO	0.01	0.009	−0.001	−0.001	0.012*** (−6.689)	0.009*** (6.653)	
UPROD	−0.008	−0.003	0.001	0.005	−0.009*** (2.971)	−0.008*** (−4.671)	
UDISX	0.004	−0.005	0	−0.007	0.004*** (−2.734)	0.003*** (3.408)	
Fund	2.431	1.306	3.317	1.446	−0.886*** (10.805)	−0.141** (−2.556)	
Dib	6.353	6.514	6.218	6.484	0.135*** (−5.44)	0.03*** (13.057)	
Lev	0.459	0.458	0.43	0.422	0.029*** (−5.844)	0.036*** (6.058)	
Size	23.152	23.051	22.346	22.2	0.806*** (−21.418)	0.85*** (22.133)	
Growth	0.336	0.095	0.294	0.1	0.042 (−0.608)	−0.006 (−0.637)	
Roe	0.082	0.084	−0.021	0.061	0.102*** (−3.859)	0.023*** (13.953)	
Top1	58.789	56.82	28.821	28.74	29.968*** (−151.117)	28.08*** (0)	
Big4	0.154	0	0.044	0	0.11*** (−12.877)	0*** (18.688)	
Notes: The numbers in parentheses are t-statistics from t-tests for the differences in means, and z-statistics from Wilcoxon tests for the differences in medians. Below is the same.

The first and second columns of Table 8 provide the relevant empirical regression results. In the “non-absolute control by major shareholders group”, the coefficient for fund holdings (Fund) is −0.007, significant at the 1 % level, indicating a significant inhibitory effect of fund holdings on REM. In the “absolute control by major shareholders group”, this coefficient decreases to −0.0038, with the significance level dropping to 10 %, indicating that when the major shareholder has higher control, the inhibitory effect of fund holdings weakens. This finding is in line with existing researches [33,34].Table 8 Regression results of equity structure heterogeneity.

Table 8	Absolute control by major shareholders	Non-Absolute control by major shareholders	State-owned enterprises	Non-State-owned enterprises	
REM	REM	REM	REM	
Fund	−0.0038* (−1.64)	−0.007*** (−18.42)	−0.004*** (−6.55)	−0.008*** (−17.02)	
Size	0.0045 (1.23)	−0.0001 (−0.07)	−0.007*** (−3.5)	−0.003 (−1.23)	
Lev	0.152*** (5.96)	0.166*** (17.92)	0.149*** (11.31)	0.167*** (14.81)	
Roe	−0.08*** (−4.18)	−0.0001 (−0.18)	−0.0005 (−0.34)	−0.0002 (−0.26)	
Growth	−0.0058*** (−3.32)	−0.0005 (−1.35)	−0.0005 (−1.27)	−0.0014* (−1.74)	
Top1	−0.0017** (−2.98)	−0.0004** (−2.61)	−0.0005** (−2.74)	−0.0013*** (−8.9)	
Big4	−0.038** (−2.89)	−0.057*** (−6.72)	−0.031*** (−3.64)	−0.07*** (−6.49)	
Hausman	53.339***	207.41***	41.95***	207.14***	
Industry and Year FE	Yes	Yes	Yes	Yes	
No. of obs	1887	11636	4604	8919	
R squared	0.073	0.179	0.124	0.218	

In summary, the role of fund holdings in curbing REM exhibits significant heterogeneity among companies with different equity balance structures, particularly being more effective in companies where the control of major shareholders is more dispersed. In other words, the degree of equity balance affects the governance effect of fund holdings on earnings management. A highly balanced equity structure enhances the governance effect of fund holdings, thereby validating the hypothesis H2 of this study.

4.3.2 Equity nature perspective

In China, state-owned enterprises and non-state-owned enterprises exhibit distinctly different operational characteristics due to their nature and objectives. State-owned enterprises typically enjoy the advantage of priority resource allocation, and profitability may not be their primary goal, thereby reducing their motivation and pressure to engage in earnings management [51,55]. In contrast, non-state-owned enterprises, due to the significance of profit objectives, may have a higher motivation for earnings management. Moreover, the unique equity structure of state-owned enterprises and the political influence behind them may weaken the supervisory effect of external overseers, such as funds, on their operational activities [36,56]. Thus, we anticipate that the inhibitory effect of fund holdings on REM will be more pronounced in non-state-owned enterprises.

To test this hypothesis, the study splits the sample into “non-state-owned enterprises” and “state-owned enterprises” and conducts empirical analyses separately for each group. Before conducting the regression analysis, we also performed a comparative analysis of the two subgroups. Table 9 reveals significant differences in both the mean and median levels of REM between the “non-state-owned enterprise group” and the “state-owned enterprise group”. Specifically, compared to non-state-owned enterprises, the mean and median REM of state-owned enterprises are higher by 0.045 and 0.027 respectively. Additionally, significant differences in mean and median values are observed for nearly all variables across the different subgroups. This suggests that our division of the sample based on equity nature is statistically significant.Table 9 Sample statistical analysis from the perspective of equity nature.

Table 9	State-owned enterprises	Non-State-owned enterprises	Difference in	
	Mean	Median	Mean	Median	Means (t-stat)	Medians (z-stat)	
REM	0.028	0.026	−0.017	0.001	0.045*** (−14.201)	0.027*** (12.483)	
UCFO	−0.005	−0.004	0.003	0.003	−0.008*** (6.262)	−0.007*** (−6.66)	
UPROD	0.017	0.015	−0.01	−0.002	0.027*** (−13.872)	0.017*** (14.136)	
UDISX	−0.007	−0.007	0.004	−0.006	−0.011*** (10.438)	−0.001*** (−5.495)	
Fund	2.882	1.422	3.354	1.414	−0.472*** (6.175)	0.008 (0.554)	
Dib	6.284	6.497	6.212	6.484	0.072*** (−3.435)	0.013*** (9.09)	
Lev	0.497	0.504	0.402	0.393	0.095*** (−25.762)	0.111*** (25.948)	
Size	23.073	22.929	22.141	22.028	0.932*** (−38.131)	0.901*** (0)	
Growth	0.295	0.075	0.303	0.117	−0.008 (0.077)	−0.042*** (−10.484)	
Roe	0.042	0.06	−0.031	0.066	0.073* (−1.864)	−0.006*** (−5.114)	
Top1	38.037	36.19	30.405	28.94	7.633*** (−28.944)	7.25*** (28.328)	
Big4	0.102	0	0.038	0	0.065*** (−13.188)	0*** (15.034)	

The results in Table 8 show that in the state-owned enterprise group, the negative correlation between fund holdings (Fund) and REM is weaker (coefficient is −0.004, significance level is 10 %), while in the non-state-owned enterprise group, this negative relationship is more significant (coefficient is −0.008, significance level is 1 %). This finding is in line with expectations and indicates that fund holdings have a greater impact on suppressing REM in non-state-owned enterprises, thereby validating the hypothesis H3 of this study. This finding is also in line with existing research [36].

4.4 Analysis of mechanism effect

4.4.1 Internal control quality and real earnings management

Table 10 presents the regression analysis results of the impact of internal control quality on REM. Specifically, the results in the first column of the regression analysis show that the regression coefficient for internal control quality is −0.005, and it is negatively correlated at the 1 % significance level. This finding indicates a clear negative relationship between higher levels of internal control quality and lower degrees of REM behavior, thus supporting the hypothesis H4a of this study: that a sound internal control system can effectively curb management's REM activities.Table 10 Regression results of internal control quality and REM.

Table 10	REM	UCFO	UPROD	UDISX	
Dib	−0.005*** (−3.93)	0.002*** (3.87)	−0.0014 (−1.76)	0.002*** (4.30)	
Size	−0.005*** (−3.41)	0.005*** (8.18)	−0.0007 (−0.8)	−0.0002 (−0.49)	
Lev	0.180*** (20.23)	−0.07*** (−20.93)	0.1*** (18.6)	−0.001*** (−3.44)	
Roe	−0.0001 (−0.25)	0.0005* (2.01)	0.0002 (0.52)	−0.0001 (−0.59)	
Growth	−0.0008* (−2.25)	−0.0003* (−2.18)	0.0007*** (3.35)	0.002*** (15.58)	
Top1	−0.0005*** (−4.26)	0.0003*** (6.29)	−0.0002*** (−3.09)	−0.000 (−0.25)	
Big4	−0.048*** (−6.69)	0.011*** (4.19)	0.023*** (−5.28)	0.014*** (5.91)	
Hausman	96.167***	135.65***	45.493***	140***	
Industry and Year FE	Yes	Yes	Yes	Yes	
No. of obs	13523	13523	13523	13523	
R squared	0.123	0.114	0.089	0.022	

Furthermore, the results in the second, third, and fourth columns of Table 10 respectively demonstrate the impact of internal control quality on abnormal operating cash flows (UCFO), abnormal expenses (UDISX), and abnormal production costs (UPROD). Specifically, the regression coefficient of internal control quality (Dib) with UCFO and UDISX is 0.002, significant at the 1 % level, while with UPROD, the coefficient is −0.0014, significant at the 10 % level. These findings suggest that high-quality internal controls significantly reduce abnormal operating cash flows and production costs while positively impacting the management of expenses.

This further demonstrates that internal control quality not only suppresses overall REM but also exerts a significant restraining influence on its various dimensions, such as operating cash flows, expenses, and production costs. Therefore, this part of the analysis not only corroborates the hypothesis H4a of this study but also details the specific impact of internal control on different REM strategies. This finding is also in line with existing studies [57,58].

4.4.2 Mediating effect of internal control quality

To validate the mediating role of internal control quality in the relationship between fund holdings and REM, this study employed a two-step regression analysis approach. First, model (11) was used to explore the relationship between fund holdings and internal control quality; subsequently, model (12) analyzed how the relationship between fund holdings and REM changes after considering internal control quality. The specific analysis results are shown in Table 11.Table 11 Regression results for the mediating effect of internal control quality.

Table 11	Dib	REM	
Fund	0.016*** (6.76)	−0.006*** (−19.21)	
Dib		−0.004** (−2.87)	
Size	0.114*** (11.94)	0.0005 (0.35)	
Lev	−1.131*** (−20.12)	0.161*** (18.31)	
Roe	0.031*** (8.06)	−0.000 (−0.13)	
Growth	−0.01*** (−4.22)	−0.0008* (−2.21)	
Top1	0.005*** (6.38)	−0.0008*** (−6.81)	
Big4	−0.021 (−0.37)	−0.05*** (−7.12)	
Hausman	63.188***	258.16***	
Industry and Year FE	Yes	Yes	
No. of obs	13523	13523	
R squared	0.289	0.118	

In the analysis presented in Table 11, the first column displays the regression results for the impact of fund holdings on internal control quality, revealing a regression coefficient of 0.016 with a positive correlation at the 1 % significance level. This indicates that as the proportion of fund holdings increases, the quality of internal controls in listed companies also improves, reflecting a clear positive impact of fund holdings on enhancing corporate governance levels.

Further, the second column of Table 11 presents the regression results of the impact of fund holdings on REM after considering internal control quality. At this point, the regression coefficient for internal control quality (Dib) is −0.004, significant at the 1 % level, while the regression coefficient for fund holdings (Fund) adjusts to −0.006, remaining significant at the 1 % level but slightly lower than when internal control quality was not considered. This suggests that internal control quality indeed plays a certain mediating role between fund holdings and REM, thereby supporting Hypothesis H4b: high-quality internal control acts as a bridge between fund holdings and the suppression of REM behavior.

In summary, the step-by-step regression analysis confirms that fund holdings enhance a company's internal control quality and highlights the mediating role of internal controls in reducing REM behaviors. This finding underscores the significance of internal control quality and the role of fund holdings in strengthening corporate governance structures. The results support the theoretical hypothesis of this study and are consistent with the findings of most prior research [[59], [60], [61]].

4.5 Robustness test

4.5.1 Changing the measurement method of REM

In order to ensure the reliability of the research findings, this study first aims to test by altering the REM measurement method. Typically, listed companies may engage in REM by offering increased discounts or reducing costs, and this conduct might be manifested in the reduction of abnormal operating cash flow or abnormal expenses. Therefore, we developed a new variable for REM (REM2 = - UCFO - UDISX) to replace the original REM variable, aligning with established literatures [44,46]. Furthermore, we also examined another traditional approach for measuring REM (REM3 = UPROD + UCFO + UDISX). The specific results are detailed in Table 12.Table 12 Robustness test for REM2 and REM3.

Table 12	REM2	REM2	REM2	REM3	REM3	REM3	
Fund	−0.0033*** (−17.62)		−0.0033*** (−17.34)	−0.0007* (−2.13)		−0.0007* (−2.215)	
Dib		−0.0039*** (−5.46)	−0.0031*** (−4.50)		0.002* (2.278)	0.0021* (2.358)	
Size	−0.002** (−2.63)	−0.0043*** (−5.09)	−0.0016* (−2.16)	0.0228*** (6.92)	0.0216*** (6.549)	0.022*** (6.647)	
Lev	0.075*** (16.45)	0.08*** (17.21)	0.071*** (15.42)	0.0685*** (6.179)	0.074*** (6.631)	0.072*** (6.46)	
Roe	−0.0004 (−1.29)	−0.0003 (−1.07)	−0.0003 (−0.97)	0.0005 (1.303)	0.0004 (1.152)	0.0005 (1.178)	
Growth	−0.0015*** (−7.98)	−0.0015*** (−8.12)	−0.0015** (−8.15)	0.0021*** (10.202)	0.002*** (10.261)	0.0022*** (10.277)	
Top1	−0.0004*** (−7.12)	−0.0002*** (−4.58)	−0.0004*** (−6.87)	−0.0002 (−0.845)	−0.0002 (−0.882)	−0.0002 (−0.936)	
Big4	−0.026*** (−7.05)	−0.025*** (−6.69)	−0.026*** (−7.07)	−0.0189 (−1.622)	−0.0188 (−1.613)	−0.0187 (−1.604)	
Hausman	321.34***	169.5***	338.52***	84.532***	83.643***	84.929***	
Industry and Year FE	Yes	Yes	Yes	Yes	Yes	Yes	
No. of obs	13523	13523	13523	13523	13523	13523	
R squared	0.133	0.109	0.139	0.023	0.023	0.024	

In Table 12, the first column shows the impact of fund holdings on the new measure of REM2. The regression analysis reveals a significant negative regression coefficient of −0.0033 for fund holdings, indicating that Fund can significantly reduce REM2. When using REM3 as the dependent variable in the fourth column, the regression coefficient remains negative and significant. This is consistent with findings from previous benchmark regression analysis in section 4.2, suggesting that fund holdings effectively suppress REM across different measurement methods.

The second column explores the relationship between internal control quality (Dib) and REM2. The results indicate a significant negative regression coefficient of −0.0039 for Dib, reinforcing the finding that higher internal control quality is linked to lower levels of REM, consistent with previous research. However, when replacing the dependent variable with REM3 in the fifth column, it's observed that Dib's coefficient is 0.002, which contradicts earlier conclusions due to special measurement methods used for REM3.

The third column investigates the impact of fund holdings on REM2 after incorporating internal control quality (Dib). The results reveal both Dib (−0.0031) and Fund (−0.0033) coefficients to be significant at a 1 % level when considering them together; this indicates that both variables continue to significantly suppress REM intensity even after factoring in internal control quality mediating effect.

It's important to note that Fund's regression coefficient in Table 12 being smaller than that in Table 6 can be attributed to how REM is calculated – as discussed earlier in section 3.2.1 – where its correlation with UPROD is positive and its correlations with UCFO and UDISX are negative; thus, making REM more reflective of actual earnings practices compared to REM2 and REM3 used here.

Overall, Table 12's results highlight how fund holdings' governance influence over real earning management holds up under various conditions.

4.5.2 Replacing control variables

As a further validation of robustness, this study employed the method of replacing control variables. Specifically, we used return on total assets (Roa) instead of return on equity (Roe) and total debt level (TL) instead of the leverage ratio (Lev) to re-conduct the fixed-effect regression analysis. The related results are presented in Table 13.Table 13 Robustness test by replacing control variables.

Table 13	REM	REM	Dib	REM	
Fund	−0.005*** (−14.14)		0.0038* (1.64)	−0.005*** (−14.18)	
Dib		−0.0029* (−2.2)		−0.0031* (−2.41)	
Size	0.0062*** (4.65)	0.0027* (−2.03)	−0.042*** (−4.84)	0.006*** (−4.55)	
TL	0.192*** (14.07)	0.207*** (15.1)	−0.454*** (−5.14)	0.193*** (14.17)	
Roa	−0.4561*** (−25.05)	−0.5177*** (−27.81)	3.588*** (30.32)	−0.468*** (−24.85)	
Growth	−0.0004 (−1.15)	−0.00003 (−0.95)	−0.0124*** (−5.48)	−0.00003 (−1.03)	
Top1	−0.00026* (−2.4)	−0.0002 (−0.19)	0.0015* (2.09)	−0.0003** (−2.44)	
Big4	−0.0456*** (−6.67)	−0.043*** (−6.34)	−0.0014 (0.974)	−0.046*** (−6.67)	
Hausman	335.11***	269.18***	94.56***	340.01***	
Industry and Year FE	Yes	Yes	Yes	Yes	
No. of obs	13523	13523	13523	13523	
R squared	0.306	0.145	0.098	0.1128	

In Table 13, the first column displays the results between the proportion of fund holdings (Fund) and REM after introducing the new control variables. At this point, the regression coefficient of the fund holdings ratio remains at −0.005, showing a negative correlation at the 1 % significance level, which is highly similar to the analysis results using the original control variables, further confirming the original conclusion that fund holdings can effectively suppress REM behavior.

The second, third, and fourth columns of Table 13 present detailed results for testing the mediating effect of internal control quality. Specifically, the third column shows a significant positive correlation between fund holdings and internal control quality (Dib), while the second column reveals a negative correlation between internal control quality and REM. After incorporating internal control quality, the impact of fund holdings on REM (the fourth column) shows that the regression coefficient of internal control quality is −0.0031, significant at the 5 % level. Additionally, the regression coefficient of fund holdings is −0.005, significant at the 1 % level, indicating that internal control quality indeed mediates the relationship between fund holdings and REM.

Although the significance of the regression results has decreased after replacing the control variables, the overall research conclusions remain consistent with the original analysis. This not only corroborates the role of fund holdings in suppressing REM but also emphasizes the importance of internal control quality. In summary, the robustness test by replacing control variables further validates the reliability and robustness of this study's results.

4.6 Results discussion

This section empirically verifies theoretical hypotheses proposed in section 2. We first employ a fixed effect model to assess the impact of fund holdings on REM. The results demonstrate that fund holdings can significantly reduce enterprises’ level of REM, which is consistent with both theory and previous studies [12,22,23]. As identified in earlier analysis, contract friction resulting from agency conflicts is the primary cause of earnings management, exacerbated by information asymmetry, making it challenging to eliminate [19]. Institutional investors hold a unique position in the capital market, equipped with both the ability and motivation to oversee corporate management. Among institutional investors, fund holdings are the most professional and active type; thus, they have an undeniable influence on corporate earnings management. This logical chain has been affirmed by numerous scholars and is also supported by this study [24,25].

Additionally, we perform a comprehensive analysis of the varying impact of fund holdings on corporate governance, taking equity structure into account. Our empirical findings reveal that the influence of fund holdings on REM varies based on levels of equity balance and nature of equity. This phenomenon is closely linked to the attributes of China's present socialist market economy [30,33,35]. Both the degree and nature of equity have an impact on primary and secondary agency issues in corporate governance [54], as confirmed by our empirical results in this section.

Furthermore, we confirm the mediating role of internal control quality in the influence of fund holdings. In addition to supervising enterprises as external supervisors and participating in corporate governance, fund holdings also work to inhibit corporate earnings management and improve the internal control quality of enterprises. This helps to mitigate the company's information asymmetry, ultimately reducing the extent of corporate earnings management [42,43].

5 Conclusion and policy recommendations

This study, using data from Chinese A-share listed companies between 2016 and 2023, examines the governance impact of fund holdings on REM behaviors, analyzes how equity structure influences this governance effect, and identifies the mediating role of internal control quality in the process. The findings are as follows: (1) Fund holdings exhibit a significant negative correlation with the level of REM, indicating that fund holdings serve as an effective external supervisory mechanism, exerting governance over management's earnings management behaviors; (2) The governance impact of fund holdings is stronger in non-state-owned enterprises and companies with more dispersed equity. This suggests that the nature of equity and the degree of equity balance are important mechanisms influencing the effectiveness of fund governance; (3) A significant negative correlation exists between internal control quality and REM, indicating that strong internal controls can effectively curb REM behaviors. Furthermore, internal control quality plays a mediating role in the relationship between fund holdings and REM.

In summary, fund holdings can play a positive supervisory role in corporate earnings management behaviors. However, a good equity structure and high-quality internal control are important conditions for the effectiveness of fund holdings’ governance. Based on these findings, the following policy recommendations are suggested: (1) Promote Long-term Investment by Fund Companies. Regulatory authorities should encourage fund companies to engage in long-term stable investments and protect their legitimate rights and interests to enhance their equity balance role in corporate governance; (2) Improve Internal Control Systems. Listed companies should strengthen the construction of internal control systems and increase governance transparency to more effectively limit REM behaviors. Regular internal control reviews and stricter penalties for violations are recommended; (3) Promote Reform and Equity Dispersion in State-owned Enterprises. Efforts should be made to advance the reform of state-owned enterprises to avoid excessive concentration of equity by controlling shareholders. Introducing strategic investors to reduce the control of state-owned capital can provide a favorable governance environment for the supervisory role of fund holdings and promote healthy long-term development of enterprises; (4) Enhance Information Disclosure Requirements. One of the reasons fund holdings may struggle to effectively intervene in corporate earnings management is information asymmetry. Regulatory authorities should require companies to disclose financial information more transparently to increase information symmetry, helping fund managers better assess corporate earnings management behaviors. Additionally, the fiduciary responsibilities of fund managers should be clarified to ensure their actions align with the long-term interests of fund holders.

The limitations of this study include the restricted data range and time span, the limited selection of variables, the lack of certainty in establishing causality, and the insufficient analysis of the heterogeneity of fund holdings. Future research should expand the data range and markets to validate the generalizability of the results; incorporate more influencing factors for comprehensive analysis; employ rigorous research designs to test causality; and deeply explore the complex mediating role of internal control quality in the relationship between fund holdings and earnings management. These directions will further enrich and improve the research on the impact of fund holdings on corporate governance.

Data availability statement

Data will be made available on reasonable request.

CRediT authorship contribution statement

Lanlan Lian: Writing – Review & Editing, Software, Data curation, Visualization, Formal analysis. Di Zhao: Writing – original draft, Methodology, Software, Investigation. Qingli Dong: Validation, Conceptualization, Supervision, Funding acquisition.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Lanlan Lian reports financial support was provided by Dalian Academy of Social Sciences (2024dlsky217 ). Qingli Dong reports financial support was provided by 10.13039/501100012226 Fundamental Research Funds for the Central Universities (DUT23RW102 ). If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

1 https://www.wind.com.cn/index.html.

2 https://finance.caixin.com/2013-10-18/100593676.html.

3 http://www.sse.com.cn/aboutus/research/research/c/3986593.pdf.

4 http://www.ic-erm.com/honor.html.

5 https://data.csmar.com/.
==== Refs
References

1 Han M. Ding A. Zhang H. Foreign ownership and earnings management Int. Rev. Econ. Finance 80 Jul. 2022 114 133 10.1016/j.iref.2022.02.074
2 Beyer A. Guttman I. Marinovic I. Earnings Management and earnings quality: theory and evidence Account. Rev. 94 4 Jul. 2019 77 101 10.2308/accr-52282
3 Yan X. Research on the accounting information disclosure of listed companies in the capital market-accounting logic based on the capital market Account. Corp. Manag. 5 5 May 2023 81 86 10.23977/acccm.2023.050509
4 Mustapha M. Che Ahmad A. Agency theory and managerial ownership: evidence from Malaysia Manag. Audit J. 26 5 Jan. 2011 419 436 10.1108/02686901111129571
5 Zhang J. Wang G. Yan C. Can foreign equity funds outperform their benchmarks? New evidence from fund-holding data for China Econ. Model. 90 2020 11 20
6 Ward C. Yin C. Zeng Y. Institutional investor monitoring motivation and the marginal value of cash J. Corp. Finance 48 Feb. 2018 49 75 10.1016/j.jcorpfin.2017.10.017
7 Jun X. Ren H. Sun P.-W. Deriving managerial skills by dissecting holding changes of mutual funds: evidence from China Pac. Basin Finance J. 68 2021 101612
8 Bebchuk L.A. Cohen A. Hirst S. The agency problems of institutional investors J. Econ. Perspect. 31 3 Aug. 2017 89 102 10.1257/jep.31.3.89
9 Fatima S. Mortimer T. Bilal M. Corporate governance failures and the role of institutional investors in Pakistan: lessons to be learnt from UK Int. J. Law Manag. 60 2 Jan. 2018 571 585 10.1108/IJLMA-10-2016-0096
10 Velte P. Which institutional investors drive corporate sustainability? A systematic literature review Bus. Strategy Environ. 32 1 2023 42 71 10.1002/bse.3117
11 Zhang J. Ye Y. Does the research meeting affect the shareholding ratio of institutional investors in listed companies? Empirical evidence from China China Finance Rev. Int. 13 4 Jan. 2021 714 733 10.1108/CFRI-08-2020-0112
12 Agnes Cheng C.S. Reitenga A. Characteristics of institutional investors and discretionary accruals Int. J. Account. Inf. Manag. 17 1 Jan. 2009 5 26 10.1108/18347640910967717
13 Boone A.L. White J.T. The effect of institutional ownership on firm transparency and information production J. Financ. Econ. 117 3 2015 508 533
14 Borochin P. Yang J. The effects of institutional investor objectives on firm valuation and governance J. Financ. Econ. 126 1 Oct. 2017 171 199 10.1016/j.jfineco.2017.06.013
15 Chen X. Lee C.-W.J. Li J. Government assisted earnings management in China J. Account. Publ. Pol. 27 3 2008 262 274
16 Ho L.-C.J. Liao Q. Taylor M. Real and accrual‐based earnings management in the pre‐and post‐IFRS periods: evidence from China J. Int. Financ. Manag. Account. 26 3 2015 294 335
17 Fama E.F. Agency problems and the theory of the firm J. Polit. Econ. 88 2 1980 288 307
18 Shleifer A. Vishny R.W. A survey of corporate governance J. Finance 52 2 1997 737 783
19 Fan J.P. Wong T.J. Corporate ownership structure and the informativeness of accounting earnings in East Asia J. Account. Econ. 33 3 2002 401 425
20 Wang M. Which types of institutional investors constrain abnormal accruals? Corp. Gov. Int. Rev. 22 1 2014 43 67 10.1111/corg.12044
21 Meng Y. Wang X. Do institutional investors have homogeneous influence on corporate social responsibility? Evidence from investor investment horizon Manag. Finance 46 3 Jan. 2019 301 322 10.1108/MF-03-2019-0121
22 Garel A. Martin-Flores J.M. Petit-Romec A. Scott A. Institutional investor distraction and earnings management J. Corp. Finance 66 Feb. 2021 101801 10.1016/j.jcorpfin.2020.101801
23 Gu X. An Z. Chen C. Li D. Do foreign institutional investors monitor opportunistic managerial behaviour? Evidence from real earnings management Account. Finance 63 1 2023 317 351 10.1111/acfi.13015
24 Sakaki H. Jackson D. Jory S. Institutional ownership stability and real earnings management Rev. Quant. Finance Account. 49 1 Jul. 2017 227 244 10.1007/s11156-016-0588-7
25 Harford J. Kecskés A. Mansi S. Do long-term investors improve corporate decision making? J. Corp. Finance 50 2018 424 452
26 Zhang L. Li B. Mutual supervision or conspiracy? The incentive effect of multiple large shareholders on audit quality requirements Int. Rev. Financ. Anal. 83 Oct. 2022 102274 10.1016/j.irfa.2022.102274
27 Chen G.-Z. Keung E.C. Corporate diversification, institutional investors and internal control quality Account. Finance 58 3 2018 751 786
28 Saleem Salem Alzoubi E. Ownership structure and earnings management: evidence from Jordan Int. J. Account. Inf. Manag. 24 2 Jan. 2016 135 161 10.1108/IJAIM-06-2015-0031
29 Ji X. Ahmed K. Lu W. The impact of corporate governance and ownership structure reforms on earnings quality in China Int. J. Account. Inf. Manag. 23 2 Jan. 2015 169 198 10.1108/IJAIM-05-2014-0035
30 Kazemian S. Sanusi Z.M. Earnings management and ownership structure Procedia Econ. Finance 31 Jan. 2015 618 624 10.1016/S2212-5671(15)01149-1
31 Masulis R.W. Wang C. Xie F. Agency problems at dual-class companies J. Finance 64 4 2009 1697 1727
32 Margaritis D. Psillaki M. Capital structure, equity ownership and firm performance J. Bank. Finance 34 3 Mar. 2010 621 632 10.1016/j.jbankfin.2009.08.023
33 Andrias K. Separations of wealth: inequality and the erosion of checks and balances Univ. Pa. J. Const. Law 18 2016 2015 419
34 Boubaker S. Sami H. Multiple large shareholders and earnings informativeness Rev. Account. Finance 10 3 Jan. 2011 246 266 10.1108/14757701111155789
35 Bushee B.J. Carter M.E. Gerakos J. Institutional investor preferences for corporate governance mechanisms J. Manag. Account. Res. 26 2 Dec. 2014 123 149 10.2308/jmar-50550
36 Kim C.-J. Efficient Management of State-Owned Enterprises: Challenges and Opportunities 2017 4 Dec. 2017 [Online]. Available: https://www.adb.org/publications/efficient-management-state-owned-enterprises-challenges-and-opportunities
37 Lenard M.J. Petruska K.A. Alam P. Yu B. Internal control weaknesses and evidence of real activities manipulation Adv. Account. 33 Jun. 2016 47 58 10.1016/j.adiac.2016.04.008
38 Chen H. Dong W. Han H. Zhou N. A comprehensive and quantitative internal control index: construction, validation, and impact Rev. Quant. Finance Account. 49 2017 337 377
39 Boulhaga M. Bouri A. Elbardan H. The effect of internal control quality on real and accrual-based earnings management: evidence from France J. Manag. Control 33 4 Dec. 2022 545 567 10.1007/s00187-022-00348-5
40 Ashbaugh‐Skaife H. Collins D.W. Kinney W.R. Jr. LaFond R. The effect of SOX internal control deficiencies and their remediation on accrual quality Account. Rev. 83 1 2008 217 250
41 Beneish M.D. Billings M.B. Hodder L.D. Internal control weaknesses and information uncertainty Account. Rev. 83 3 2008 665 703
42 Järvinen T. Myllymäki E.-R. Real earnings management before and after reporting SOX 404 material weaknesses Account. Horiz. 30 1 2016 119 141
43 Wali S. Masmoudi S.M. Internal control and real earnings management in the French context J. Financ. Report. Account. 18 2 Jan. 2020 363 387 10.1108/JFRA-09-2019-0117
44 Cohen D.A. Zarowin P. Accrual-based and real earnings management activities around seasoned equity offerings J. Account. Econ. 50 1 2010 2 19
45 Roychowdhury S. Earnings management through real activities manipulation J. Account. Econ. 42 3 2006 335 370
46 Baatour K. Ben Othman H. Hussainey K. The effect of multiple directorships on real and accrual-based earnings management: evidence from Saudi listed firms Account. Res. J. 30 4 Jan. 2017 395 412 10.1108/ARJ-06-2015-0081
47 Zhao Q. Hu C. Ma R. Yin Y. Liu Y. Foreign shareholders and executive compensation stickiness: evidence from China Finance Res. Lett. 59 2024 104790
48 Chen D. Li J. Liang S. Wang G. Macroeconomic control, political costs and earnings management: evidence from Chinese listed real estate companies China J. Account. Res. 4 3 2011 91 106
49 Chung R. Firth M. Kim J.-B. Institutional monitoring and opportunistic earnings management J. Corp. Finance 8 1 2002 29 48
50 Warfield T.D. Wild J.J. Wild K.L. Managerial ownership, accounting choices, and informativeness of earnings J. Account. Econ. 20 1 1995 61 91
51 Dong N. Wang F. Zhang J. Zhou J. Ownership structure and real earnings management: evidence from China J. Account. Publ. Pol. 39 3 2020 106733
52 Lo A.W. Wong R.M. Firth M. Can corporate governance deter management from manipulating earnings? Evidence from related-party sales transactions in China J. Corp. Finance 16 2 2010 225 235
53 Jian M. Wong T.J. Propping through related party transactions Rev. Account. Stud. 15 2010 70 105
54 Zhang M. Gao S. Guan X. Jiang F. Controlling shareholder-manager collusion and tunneling: evidence from China Corp. Gov. Int. Rev. 22 6 2014 440 459 10.1111/corg.12081
55 Li W. Zhang R. Corporate social responsibility, ownership structure, and political interference: evidence from China J. Bus. Ethics 96 2010 631 645
56 Gourevitch P.A. Shinn J. Political Power and Corporate Control: the New Global Politics of Corporate Governance 2010 Princeton University Press
57 Cheng Q. Lee J. Shevlin T. Internal governance and real earnings management Account. Rev. 91 4 Jul. 2016 1051 1085 10.2308/accr-51275
58 Garg M. The effect of internal control certification regulatory changes on real and accrual-based earnings management Eur. Account. Rev. 27 5 Oct. 2018 817 844 10.1080/09638180.2018.1454336
59 Alhadab M. Clacher I. The impact of audit quality on real and accrual earnings management around IPOs Br. Account. Rev. 50 4 Jun. 2018 442 461 10.1016/j.bar.2017.12.003
60 Guo J. Huang P. Zhang Y. Zhou N. Foreign ownership and real earnings management: evidence from Japan J. Int. Account. Res. 14 2 Sep. 2015 185 213 10.2308/jiar-51274
61 Prawitt D.F. Smith J.L. Wood D.A. Internal audit quality and earnings management Account. Rev. 84 4 Jul. 2009 1255 1280 10.2308/accr.2009.84.4.1255
