Week 6 assignment

Mandatory internal control and earnings management Zilong Songa**, Ying Liua*,**, and Jiaxin Wang b

aSchool of Accounting, Nanjing University of Finance and Economics, Nanjing, China; bSchool of Accounting, Zhongnan University of Economics and Law, Wuhan, China

ABSTRACT This paper examines whether the mandatory implementation of an internal control system influences listed firms’ earnings management in China. We use firms that are mandated for implementing internal control as the treat- ment group, and firms that do not implement internal control as the control group. We find that the mandatory implementation of an internal control system results in an increase in both real and accrual-based earnings man- agement. We also find that the extra institutional compliance cost induced by implementing a mandatory internal control system was the main cause of the increase in earnings management. Our conclusions thus provide empiri- cal evidence that differs from that on the developed market regarding the research about mandatory internal control and earnings management.

KEYWORDS Mandatory internal control system; accrual-based earnings management; real earnings management

JEL G38; D82; M41

1. Introduction

Along with the enactment of the Sarbanes-Oxley Act in the United States in 2002, mandatory internal control systems became the worldwide mainstream in corporate internal control practices. Under the influence of globalization, China’s internal control system has undergone a transition from voluntary to mandatory implementation. To fully implement an internal control system in the Chinese listed firms, the Ministry of Finance (MoF) of China and the China Securities Regulatory Commission (CSRC) jointly formulated the “Announcement on Implementing Internal Control System for the Main Board Listed Companies by Categories and Batches” (hereafter referred to as the “Announcement”). The “Announcement,” stipulates that all listed companies on the Main Board should establish an internal control system from 2012 onward. However, given the costly expenses involved in the construction of an internal control system, and the differences in the nature of the property rights and the economic basis of the listed companies in China, the government decided to implement the mandatory internal control system by categories and batches.1

While the promulgation of the “Announcement” undoubtedly had a critically positive effect on the listed firms’ implementation of an internal control system, it is noteworthy that, against this backdrop, the levels of the listed firms’ earnings management increased during the same period. We find that the means and the medians of both real and accrual-based earnings management of the listed companies increased between 2009 and 2014. We wonder whether this earnings management increase was really caused by the promulgation of the “Announcement.” If so, what was the channel through which the “Announcement” influences the firms’ earnings management?

Traditional earnings management theories contend that, in a stringent regulatory environment, there is a substitutional relationship between real and accrual-based earnings management (Zang 2012). However, when it comes to the effects of the policies that might influence the listed firms’

CONTACT Jiaxin Wang [email protected] School of Accounting, Zhongnan University of Economics and Law, Building, No. 182, Nanhu Avenue, Donghu High-tech Development Zone, Wuhan City, Hubei Province 430073, China *Ying Liu is co-first author. **These authors equally contributed to this work.

EMERGING MARKETS FINANCE AND TRADE 2022, VOL. 58, NO. 12, 3439–3453 https://doi.org/10.1080/1540496X.2022.2050903

© 2022 Taylor & Francis Group, LLC

earnings management strategies, the traditional theories do not apply to the Chinese context. This is because mandatory internal control could impose heavy institutional compliance costs upon firms, which is confirmed by a substantial amount of research (Kinney and Shepardson 2011). The excessive compliance costs might impose excessive cost pressure on firms (Iliev 2010), and cause firms to change their earnings management strategies. This is especially factual in China, as the Chinese economy is currently undergoing a structural adjustment and thus its market competition is extremely intense, which strongly stimulates firms’ incentives to pursue financial performance.

This study explores the effect of the implementation of a mandatory internal control system on the firms’ earnings management in the Chinese context. Taking the promulgation of the “Announcement” as the exogenous event, we test this effect using a difference-in-difference (DID) method and a unique panel data set from 6,981 publicly traded A-share main board firm-year observations for the years 2009–2014 in China. We find that the promulgation of the “Announcement” indeed resulted in the increase in both real and accrual-based earnings management. We also find that the extra institutional compliance cost induced by implementing a mandatory internal control system was the main cause of the increase in earnings management.

This study contributes to related literature in three ways. First, this paper extends the internal control literature by enriching the research regarding the economic consequences of mandatory internal control. Second, this study complements the earnings management literature that focuses on the incentives of firms to evade monitoring and on the substitutional relationship between real and accrual-based earnings management. Third, by analyzing the peculiarities of emerging economies, this paper adds to the literature regarding public policy assessment. Due to the differences between China and Western countries, the internal control system formulated in the “Announcement” in China is quite different from that formulated under the framework of the SOX Act of the United States. In particular, the mandatory internal control system in China was implemented by categories and batches, which makes the research setting more interesting and distinct. These differences make the implementation mode and the economic consequences of the Chinese internal control system different from those in the US. We are thus the first to find that the mandatory internal control system in China, instead of acquiring its intended effects on improving earnings quality, leads to firms’ distorted reporting behavior and a lower earnings quality. Therefore, it is more accurate and practical to consider the context-specific peculiarities when assessing such mandated public policies.

The remainder of this paper is structured as follows: First, we present a review of related literature and develop hypothesis. Next, we describe our research method and present the results of our analyses. Finally, we provide conclusions and discuss implications.

2. Literature and Hypothesis

The extant research on mandatory internal control focuses on the issue of the cost-benefit induced by the system. Iliev (2010), for example, documents a positive relationship between the SOX implemen- tation and earnings quality. However, since implementing the SOX regulations requires a substantial number of expenses, primarily including the control system development and maintenance, system implementation and supervision, and internal control audit fees, etc., others argue that a common problem of mandatory internal control is how to balance the relationship between implementation costs and the enterprise development (Iliev 2010). Moreover, Kinney and Shepardson (2011) point out that the cost burden caused by the SOX Section 404 has a greater impact on small companies. Consequently, the issue of cost-benefit as it relates to mandatory internal control remains unsolved.

In the sense that the implementation criteria have become more standardized, and the scope of the implementation has been expanded, the nature of China’s internal control system has been trans- formed from an inducing to a compulsory system. This is reflected in the transition from the voluntary implementation encouraged by the “Basic Standards for Corporate Internal Control” (hereafter: the “Standards”) in 2008 to the mandatory implementation stipulated in the “Announcement” in 2012. The supervision and penalties for inadequate implementation have become simultaneously stricter.

3440 Z. SONG ET AL.

This transitional process could be justified by the mandatory institutional transition hypothesis in the institutional economic theory. Specifically, the government’s intervention into policy implementation could mitigate the problem of insufficient institutional supply caused by an institutional transition that is inducing, albeit not regulatory (Lin 1989).

However, such governmental intervention might have opposite effects. One of such effects is the excessive costs caused by mandated policy implementation. As Iliev (2010) points out, “Section 404 costs and regulatory burden are far beyond what Congress intended and well in excess of the benefits to shareholders and management,” a mandatory institutional arrangement does impose excessive costs to firms. Related studies additionally confirm that mandatory internal control causes substantial institu- tional compliance costs. The 2004 Financial Executives International Survey showed, for example, that the average compliance cost of the SOX Act of the American listed firms was approximately USD 4.36 million per year,2 which dramatically exceeded the USD 0.91 million3 that was originally estimated by the US Securities and Exchange Commission (SEC). Other research indicates that the SOX Section 404 increased firms’ audit fees significantly (Kinney and Shepardson 2011). In addition, Kinney and Shepardson (2011) find that the implementation of the SOX Section 404 (b) imposes a heavier burden of audit costs on small companies, which seriously hinders the growth of such enterprises.

As a mandatory institutional arrangement, the “Announcement” could bring about extra institu- tional compliance costs (Iliev 2010), including explicit and hidden costs. The “explicit cost” refers to the actual expenses that firms incur by implementing a mandatory internal control system; this includes the cost of designing and implementing the system, such as the fees paid to internal control consulting organizations, the operation and maintenance costs of an information system, and the labor costs of relevant personnel. Internal control audit and information disclosure costs are impor- tant constituents of the explicit costs as well (Maher and Weiss 2010). The “hidden cost” refers to the opportunity cost of a firms’ loss; this includes such things as the decision-making mistakes caused by an increasingly complicated process, the decrease in employee initiative and work enthusiasm caused by over-monitoring, and the decline in work efficiency caused by employees’ resistance to strict control. Although difficult to measure by accounting numbers, these hidden costs have a negative impact on corporate performance.

Specifically, the institutional compliance costs are reflected in the internal control implementation process. On the one hand, although mandatory internal control helps to improve business manage- ment security, the management is more concerned about the extra costs, including the service fees for purchasing technology, internal control audit fees, and the implementation and maintenance fees of the system. These costs typically require additional budget for which it is difficult to generate significant benefits in the short term, and thus would reduce managerial performance. On the other hand, from the employee perspective, many of them have some opposition to the mandatory internal control system. Since internal control requires strict approval and monitoring of the employees’ operations, it typically reduces work satisfaction, leads to a decline in labor productivity, and thus increases the hidden costs for the system. The employees may additionally consider the imposed business management security to be contradictory to the convenience of routine work. This is particularly true in transitional economies, where the suspicion of internal control exists overwhel- mingly in the employees’ mind-set.

Given the deleterious effects, the extra institutional compliance costs would be an essential concern of the management. Since extra costs could undermine corporate financial performance and thus managerial performance, the performance-sensitive and self-serving managers would attempt to compensate for the explicit and potential losses (Chung, Firth, and Kim 2002) caused by implementing a mandatory internal control system. From this perspective, earnings management would be a viable option that is extremely likely to be used by managers. After the promulgation of the “Announcement,” the increase in institutional compliance costs would severely reduce firms’ profits, imposing excessively higher performance pressure on the management. To meet or beat their corporate financial targets, to secure managerial performance remuneration, and to increase the

EMERGING MARKETS FINANCE AND TRADE 3441

probability of exercising their options (Bartov, Givoly, and Hayn 2002), the managers may use the two types of earnings management simultaneously (Zhu et al. 2015). Qin, Cai, and Wei (2021) find that managers manipulate the earnings management when facing mandatory contributions, for example, FASB mandates disclosure of pension asset composition and a description of investment strategy under SFAS 132 R. In addition, existing research points out that due to the relative lack of institutional investor supervision, it is difficult for external investors in China to detect earnings management behaviors (Gao et al. 2020). Therefore, we anticipate that the promulgation of the “Announcement” would lead to an increase in both real and accrual-based earnings management. Based on the above analysis, we propose the following hypothesis:

H: Compared to voluntary implementation, the mandatory implementation of an internal control system is more likely to increase both accrual-based earnings management and real earnings management.

3. Research Design

3.1. Sample and Data

To test our conjecture empirically, we take the promulgation of the “Announcement” in 2012 as the benchmark and use all of the A-share listed firms on the main board in Shanghai and Shenzhen Stock Exchanges in 2009–2014 as the sample. The years 2009–2014 are selected as the sample interval because China conducted a major reform of the internal control system of enterprises in 2008 and implemented it in the following year.4 According to the “Announcement,” the year 2014 is designated as the completion period because all the firms listed on the Main Board should have their mandatory internal control construction completed by then. Our main data source is the CSMAR database, which contains both financial and non-financial information of the Chinese listed firms. The initial sample contains all of the data for the three years before and after the promulgation of the “Announcement.” The sample is filtered according to the following procedures: 1) We exclude the firms that belong to the financial and insurance industries, and those that have been marked ST, *ST during our sample period. 2) We exclude the firms with missing financial data. 3) To avoid the influence of extreme outliers, we winsorize all of the continuous firm-level variables at the 1% and 99% levels. Using these procedures, we obtain 6,977 observations.

3.2. Methods and Variables

3.2.1. Accrual-based Earnings Management Models Drawing on Dechow, Sloan, and Sweeney (1995), we use cross-sectional Jones Model to calculate discretionary accruals. We do OLS regressions by years and industries with Models (1), and use the estimations to calculate the discretionary accruals with Model (2) and (3).

TACi;t

TAi;t� 1 ¼ β1

1 TAi;t� 1

þ β2 ΔSalesi;t

TAi;t� 1 þ β3

PPEi;t

TAi;t� 1 þ εi;t (1)

NDAi;t

TAi;t� 1 ¼ β1

1 TAi;t� 1

þ β2 ΔSalesi;t � ΔARi;t

TAi;t� 1 þ β3

PPEi;t

TAi;t� 1 (2)

DAi;t ¼ TACi;t

TAi;t� 1 �

NDAi;t

TAi;t� 1 (3)

3442 Z. SONG ET AL.

where DA represents discretionary accruals. The higher DA is, the more accrual-based earnings management there is. TAC is total accruals, computed as net profit minus net operating cash flows. TA is total assets. Sales is the change in net sales. PPE is the original value of fixed assets. NDA is non- discretionary accruals. AR is the change in accounts receivable.

3.2.2. Real Earnings Management Models Consistent with existing literature (Cohen, Dey, and Lys 2008; Gunny 2010; Shen et al. 2020; Xu and Rao 2021; Zang 2012), we use abnormal production costs, abnormal discretionary expenses, and abnormal operation cash flows to estimate real earnings management. We do OLS estima- tion by years and industries with Models (4) to (6), respectively. The abnormal manipulations, including abnormal production costs (AB PROD ROA), abnormal discretionary expenses (AB DISX ROA) and abnormal operation cash flows (AB CFO ROA), are calculated as the actual values minus their estimations.

PRODi;t

TAi;t� 1 ¼ β0 þ β1

1 TAi;t� 1

þ β2 Salesi;t

TAi;t� 1 þ β3

Salesi;t

TAi;t� 1 þ β4

ΔSalesi;t� 1

TAi;t� 1 þ εi;t (4)

DISXi;t

TAi;t� 1 ¼ β0 þ β1

1 TAi;t� 1

þ β2 Salesi;t� 1

TAi;t� 1 þ εi;t (5)

CFOi;t

TAi;t� 1 ¼ β0 þ β1

1 TAi;t� 1

þ β2 Salesi;t

TAi;t� 1 þ β3

ΔSalesi;t

TAi;t� 1 þ εi;t (6)

where PROD represents total production cost, which equals to cost of goods sold and the change in inventories. DISX means discretionary expenses, computed as the sum of selling, general, and administrative expenses (SG&A). CFO means net operation cash flows. TA is firm’s total assets. S is net sales. S is the change in net sales. ROA is return on assets.

In accordance with Zang (2012), we then calculate the indicator for real earnings management (REM ROA) with the following model:

REM ROA ¼ AB PROD ROA � AB DISX ROA � AB OCF ROA (7)

where the larger REM ROA is, the more real earnings management there is.

3.2.3. Difference-in-difference (DID) Model We analyze our research questions using a dynamic DID model. Unlike a static DID model, which applies to situations where a policy dummy variable only occurs once, a dynamic DID model is more suitable to situations where a policy dummy variable occurs multiple times. As noted above, the “Announcement” implements the mandatory internal control system by cate- gories and batches. Therefore, a dynamic DID model is a better option for analyzing the effects of the “Announcement.” While the premise of the DID method is the random selection of both the experimental and the control groups, the unmatched firms from both groups may have certain differences in the level of earnings management per se, which might influence the robustness of our conclusion. In considering this issue, we do propensity score matching (PSM) for the samples between the experimental and the control groups. Since the samples in the experimental group with PSM are easier to match with those that have similar characteristics in the control group, the PSM-DID method could effectively deal with the design of intertem- poral quasi-natural experiments, thus better revealing the intrinsic characteristics of mandatory implementation by categories and batches (Gong et al. 2016). To be precise, we use all of the control variables included in the main regression models as characteristic variables, and we

EMERGING MARKETS FINANCE AND TRADE 3443

adopt probit regressions to conduct one-to-one matching with the estimations. To test the relationship between the mandatory implementation and real earnings management, we design the following DID model:

EMi;t ¼ β0 þ β1AFTERi;t þ β2TREATi;t þ β3AFTERi;t � TREATi;t þ YEARþ INDUSTRY þ γXi;t þ εi;t

(8)

where EM represents earnings management, proxied by accrual-based earnings management (DA) and real earnings management (REM ROA). We also include three methods of real activities manip- ulation (AB PROD ROA, AB DISX ROA and AB CFO ROA) as robustness checks. AFTER is a dummy, taking the value 1 after the policy implementation, and 0 in the years before the policy. TREAT is to differentiate the treatment and the control groups, taking the value 1 if a firm belongs to the treatment group, and 0 if a firm belongs to the control group. X represents the other control variables. The definitions of all of the variables are displayed in Table 1.

4. Empirical Results

4.1. Descriptive Statistics

The descriptive statistics of the main variables are presented in Table 2. The standard deviations of the accrual-based earnings management (0.104) and the real earnings management (0.264) are relatively high, meaning that there is a big difference in the earnings management among the Chinese listed firms. In addition, the mean of LEVERAGE (0.514) indicates that the leverage ratios of the Chinese listed firms are relatively high. The means of LASSET (22.36) and ROA (0.043) indicate that the average scale of the assets and profitability of the Chinese listed firms is not very ideal. Moreover, the standard deviations of BM; TENURE; LASSET and CR are all quite high, indicating high fluctuations of book-to-market ratios, auditors’ terms of service, firm scales and current ratios among the Chinese listed firms.

Table 1. Variable definitions.

Variables Definition

Variables employed in the DID Model DA Accrual-based earnings management, calculated with Model (3). AB PROD ROA ROA-adjusted abnormal production costs, calculated with Model (4). AB DISX ROA ROA-adjusted abnormal discretionary expenses, calculated with Model (5). AB OCF ROA ROA-adjusted abnormal operation cash flows, calculated with Model (6). REM ROA Composite indicator of real earnings management, calculated with Model (7). TREAT Group differentiation dummy, taking the value 1 if a firm has implemented internal control mandatorily, and 0

otherwise. AFTER Implementation year dummy, taking the value 1 in the mandatory implementation year and the years after, and 0

otherwise. GROWTH The growth rate of operation income. BM Book to market value. LNLAG Audit delay, computed as the natural logarithm of the length between January 1 and audit report signing date. TENURE The length of auditor’s term of service. LASSET The natural logarithm of total assets. INVREC Computed as the sum of inventory and accounts receivable divided by average total assets. ROA Return on assets, computed as net profit divided by average total assets. LEVERAGE Liability ratio, computed as total liabilities divided by total assets. CR Current ratio, computed as total current assets divided by total current liabilities. AGEEST Firm age, proxied by firm’s listing length. BOAIND Proportion of independent directors on board.

3444 Z. SONG ET AL.

4.2. Empirical Results

The results of the DID test for the relationship between the mandatory internal control and the firms’ accrual-based earnings management are reported in Panel A of Table 3. The results of Column (1) show that DA is significantly and positively correlated with TREAT � AFTER, meaning that, compared with that of the firms that have not implemented internal control, the extent of the accrual-based earnings management of the firms that implemented the internal control mandatorily increased significantly after the promulgation of the “Announcement.” Columns (2) and (3) display the results of the tests conducted in the two groups with either positive or negative accrual-based earnings management, respectively. The coefficient between DA and TREAT � AFTER is significantly positive in the group with positive accrual-based earnings management, but insignificant in the group with negative accrual-based earnings management. These results show that, after the implementation of a mandatory internal control system, the management is more inclined to manipulate the earnings upward. This finding also confirms the channel of institutional compliance cost through which the mandatory internal control influences firms’ earnings management.

The results of the DID test for the relationship between the mandatory internal control and the firms’ real earnings management are reported in Panel B of Table 3. Column (1) displays the result of the test with the composite indicator of the real earnings management. REM ROA is significantly and positively correlated with TREAT � AFTER, meaning that, the extent of the real earnings manage- ment of the firms in the experimental group increased significantly after the promulgation of the “Announcement.” Columns (2) to (4) show the results of the three types of real activities manipula- tion. AB PROD ROA is positively and significantly correlated with TREAT � AFTER, indicating that, the firms in the treatment group tended to increase earnings through over-production and scale effect, leading to higher total production costs. The coefficient between AB DISX ROA and TREAT � AFTER is insignificant, indicating that the influence of the mandatory internal control implementa- tion on discretionary expenses is trivial. AB OCF ROA is negatively and significantly correlated with TREAT � AFTER, suggesting that, after the promulgation of the “Announcement,” the firms in the treatment group tended to increase earnings through relaxed credit policies, leading to lower opera- tion cash flows.

In summary, the results show that both the accrual-based and the real earnings management increased significantly after the implementation of a mandatory internal control system. This finding thus verifies our research hypothesis.

Table 2. Descriptive statistics of the main variables.

Variables N Mean SD Min P25 P50 P75 Max

DA 6977 0.0199 0.1036 −0.3402 −0.0297 0.0144 0.0630 0.4744 AB PROD ROA 6754 −0.0704 0.1483 −0.6049 −0.1328 −0.0610 −0.0020 0.4506 AB DISX ROA 6754 0.0362 0.0779 −0.1760 −0.0024 0.0220 0.0612 0.3569 AB OCF ROA 6754 0.0264 0.1010 −0.3657 −0.0200 0.0281 0.0793 0.3412 REM ROA 6754 −0.1320 0.2638 −1.0651 −0.2416 −0.1119 −0.0071 0.7625 GROWTH 6977 0.1972 0.5140 −0.6041 −0.0171 0.1130 0.2690 3.8906 BM 6977 1.1346 1.0174 0.0940 0.4651 0.7902 1.4250 4.9835 LNLAG 6977 4.4793 0.2579 3.3673 4.3820 4.4886 4.6821 4.7875 TENURE 6977 5.1316 3.8042 1 2 4 7 16 LASSET 6977 22.3587 1.2928 18.9563 21.4734 22.1905 23.1483 25.7349 INVREC 6977 0.3481 0.3260 0.0000 0.1287 0.2642 0.4396 1.5253 ROA 6977 0.0430 0.0592 −0.2025 0.0137 0.0365 0.0689 0.2517 LEVERAGE 6977 0.5143 0.2059 0.0395 0.3647 0.5212 0.6668 1.0974 CR 6977 1.7696 1.9395 0.2064 0.9124 1.3077 1.9331 23.7967 AGEEST 6977 2.7115 0.3456 1.0986 2.5649 2.7726 2.9444 3.2958 BOAIND 6977 0.3685 0.0537 0.3077 0.3333 0.3333 0.3846 0.5714

EMERGING MARKETS FINANCE AND TRADE 3445

Table 3. Mandatory internal control and firms’ earnings management.

Panel A: Mandatory Internal Control and Firms’ Accrual-based Earnings Management

Variables

(1) (2) (3)

DA Positive DA Negative DA

TREAT −0.016*** −0.014** 0.001 (−2.71) (−2.54) (0.13)

AFTER 0.003 0.008 0.014** (0.34) (1.08) (2.06)

TREAT � AFTER 0.012* 0.011* −0.004 (1.87) (1.80) (−0.70)

GROWTH −0.000 0.035*** −0.035*** (−0.00) (7.40) (−8.34)

BM 0.004* −0.003 0.011*** (1.90) (−1.30) (5.71)

LNLAG −0.001 0.002 −0.008 (−0.23) (0.31) (−1.53)

TENURE −0.000 −0.000 −0.000 (−0.94) (−0.97) (−0.31)

LASSET 0.004** −0.005*** 0.007*** (2.10) (−2.98) (5.12)

INVREC 0.043*** 0.031*** 0.008 (5.28) (3.66) (1.16)

ROA 0.431*** 0.261*** 0.234*** (11.01) (6.44) (7.68)

LEVERAGE 0.004 0.074*** −0.067*** (0.36) (5.01) (−6.39)

CR 0.002** 0.002** −0.000 (2.11) (2.52) (−0.81)

AGEEST 0.001 0.009* −0.003 (0.20) (1.90) (−0.56)

BOAIND −0.002 0.023 −0.009 (−0.07) (0.85) (−0.40)

YEAR Yes Yes Yes INDUSTRY Yes Yes Yes Constant −0.102** 0.099** −0.164***

(−2.25) (2.21) (−4.30) N 6977 4122 2855 R2 0.129 0.220 0.245

Panel B: Mandatory Internal Control and Firms’ Real Earnings Management

Variables

(1) (2) (3) (4)

REM ROA AB PROD ROA AB DISX ROA AB OCF ROA

TREAT −0.058*** −0.030*** 0.017*** 0.011* (−3.38) (−3.43) (2.71) (1.95)

AFTER 0.051*** 0.025** −0.019*** −0.009 (2.74) (2.34) (−3.03) (−1.26)

TREAT � AER 0.048*** 0.017** −0.016*** −0.014** (3.35) (2.23) (−3.75) (−2.39)

GROWTH 0.013 0.006 0.009*** −0.007 (1.05) (0.86) (2.82) (−1.41)

BM 0.017*** 0.011*** −0.007*** 0.000 (3.15) (3.76) (−3.74) (0.12)

LNLAG −0.004 −0.006 0.004 −0.006 (−0.24) (−0.83) (0.94) (−1.05)

TENURE −0.001 −0.001 0.000 0.000 (−0.84) (−1.04) (0.34) (0.38)

LASSET 0.015*** 0.009*** −0.006*** 0.001 (2.97) (3.44) (−3.39) (0.68)

INVREC 0.200*** 0.124*** −0.012* −0.061*** (9.22) (10.59) (−1.65) (−8.85)

ROA −1.461*** −0.864*** 0.162*** 0.433*** (−13.58) (−14.76) (4.28) (12.75)

LEVERAGE −0.026 −0.028 0.012 −0.014 (−0.74) (−1.60) (1.03) (−1.18)

(Continued)

3446 Z. SONG ET AL.

4.3. Robustness Checks

4.3.1. Stage-by-stage Test We create three dummy variables, where Post_1 indicates that the sample year is one year before the enforcement of internal control; Post1 indicates that the sample year is the year when the internal control is implemented; Post1+ indicates that the sample year is all years after the implementation of internal control. By multiplying these three dummy variables with the dummy variable TREAT, we observe the dynamic effects of the mandatory implementation of internal control on earnings quality.

Table 4 displays the results. We find that the coefficients of TREAT � Post_1 are not significant. This indicates that before the mandatory implementation of internal control, there is no significant difference in earnings quality between the treatment group and the control group, for both accrual- based earnings management and real earnings management. But the coefficients of TREAT � Post1

Table 3. (Continued).

Panel B: Mandatory Internal Control and Firms’ Real Earnings Management

Variables

(1) (2) (3) (4)

REM ROA AB PROD ROA AB DISX ROA AB OCF ROA

CR 0.008*** 0.004** −0.001 −0.003*** (3.04) (2.53) (−0.77) (−3.32)

AGEEST 0.014 0.011 −0.003 0.002 (0.96) (1.43) (−0.58) (0.34)

BOAIND −0.082 −0.043 0.037 −0.017 (−0.97) (−0.97) (1.14) (−0.64)

YEAR Yes Yes Yes Yes INDUSTRY Yes Yes Yes Yes Constant −0.415*** −0.214*** 0.126*** 0.059

(−3.11) (−3.06) (2.66) (1.32) N 6754 6754 6754 6754 R2 0.249 0.274 0.139 0.170

***, ** and * indicate statistical significance at 1%, 5%, and 10% levels respectively.

Table 4. Stage-by-stage test.

Variables

(1) (2) (3) (4) (5)

DA REM ROA AB PROD ROA AB DISX ROA AB OCF ROA

TREAT −0.017*** −0.044** −0.204*** 0.014** 0.026** (−2.60) (−2.27) (−4.68) (2.05) (2.19)

Post_1 −0.008 0.016 0.082 −0.016** −0.037 (−0.87) (0.59) (0.79) (−1.97) (−1.36)

Post1 −0.002 0.057 −0.229 −0.034*** −0.006 (−0.15) (1.55) (−1.64) (−2.60) (−0.24)

Post1+ 0.002 0.102** −0.704*** −0.053*** −0.014 (0.11) (2.12) (−4.62) (−3.03) (−0.52)

TREAT � Post 1 0.005 −0.010 0.028 −0.006 0.002 (0.70) (−0.55) (0.40) (−1.30) (0.11)

TREAT � Post1 0.013* 0.035* 0.255*** −0.014*** −0.054*** (1.74) (1.90) (3.20) (−2.59) (−3.39)

TREAT � Post1þ 0.017** 0.029* 0.294*** −0.009* −0.051*** (2.25) (1.65) (4.66) (−1.75) (−3.57)

Control variables Yes Yes Yes Yes Yes YEAR Yes Yes Yes Yes Yes INDUSTRY Yes Yes Yes Yes Yes N 6977 6754 6754 6754 6754 R2 6977 6754 6754 6754 6754

***, ** and * indicate statistical significance at 1%, 5%, and 10% levels respectively.

EMERGING MARKETS FINANCE AND TRADE 3447

and TREAT � Post1+ are significant, and carry the right signs, indicating that after the mandatory implementation of internal control, the earnings quality of the treatment group is significantly worse than that of the control group. This finding verifies our main results.

4.3.2. Placebo Test In line with Tan et al. (2015), we conduct placebo tests by pushing the policy execution time forward. If the coefficient of the interaction term becomes insignificant after changing the policy execution time, this would indicate that the stimulating effect on the firms’ earnings manage- ment comes from the exogenous event of the “Announcement.” Based on this perspective, we choose 2010 as the placebo exogenous event and we re-do the DID test with Model (8). These results are reported in Table 5. We find that the coefficient of the interaction term TREAT � AFTER is insignificant, which confirms that it is indeed the exogenous event of the “Announcement,” rather than other policy changes or random factors, that increased the firms’ earnings management.

4.3.3. Other Robustness Tests In addition to the above-mentioned tests, we conduct the following robustness tests. Because of space limitations, the results are not reported.

First, we do the DID test again with newly constructed dependent variables. Consistent with Cohen, Dey, and Lys (2008), we re-estimate the composite indicator of real earnings management (REM). The results show that the newly estimated indicators of real earnings management are still positively and significantly correlated with the interaction term TREAT � AFTER, thus demonstrating the robust- ness of our previous conclusion.

Second, we conduct quantile regressions with 10, 25, 50, 75 and 90% percentiles of the earnings management indicators. We find that the positive correlation between the earnings management indicators and TREAT � AFTER displays a decreasing trend as the percentiles get lower, which shows the robustness of our conclusion.

Third, we estimate the nonparametric local linear regression coefficients for the three bandwidths of Lwald100, Lwald50 and Lwald150. We find that the estimated coefficients of the triangular and rectangular cores of the earnings management indicator are generally significantly positive, thus verifying our conclusion.

Table 5. Placebo test.

Variables

(1) (2) (3) (4) (5)

DA REM ROA AB PROD ROA AB DISX ROA AB OCF ROA

TREAT −0.017** −0.035* −0.023** 0.012* 0.004 (−2.36) (−1.77) (−2.22) (1.68) (0.60)

AFTER −0.030*** −0.003 −0.003 −0.017** 0.022** (−2.88) (−0.13) (−0.18) (−2.41) (2.20)

TREAT � AFTER 0.011 0.007 −0.001 −0.007 −0.005 (1.53) (0.45) (−0.11) (−1.45) (−0.61)

Control variables Yes Yes Yes Yes Yes YEAR Yes Yes Yes Yes Yes INDUSTRY Yes Yes Yes Yes Yes N 5632 5415 5415 5415 5415 R2 0.127 0.264 0.267 0.163 0.186

***, ** and * indicate statistical significance at 1%, 5%, and 10% levels respectively.

3448 Z. SONG ET AL.

Finally, we randomly select the same number of samples from the original regression model as a new experimental group and the remaining samples as a new control group. We do the DID test again with the new groups and we repeat the simulations 5,000 times. The results show that the coefficients of TREAT � AFTER are all insignificant, indicating that the main regression results are not driven by accidental factors; therefore, the conclusion is robust.

5. Extensions

5.1. Channel

As discussed above, the institutional compliance cost induced by mandatory internal control is the main factor that stimulates the firms’ earnings management. Under this conceptual frame- work, we can identify the channel from the following two aspects: First, the implementation of a mandatory internal control system is very likely to cause the institutional compliance costs to increase. Second, the extra compliance cost pressure spurs the management to use more earnings management.

To verify the first aspect of the channel, we replace the dependent variable of Model (8) with the institutional compliance cost, proxied by three variables (Inter_Audit_Fee, Mana_fee and Audit_fee). Inter_Audit_Fee indicates the natural logarithm of the internal control audit fees paid by listed

Table 6. Channel.

Panel A: The Influence of Mandatory Internal Control on Institutional Compliance Cost

Variables

(1) (2) (3) (4) (5)

Institutional compliance cost Internal control quality

Classification shifting

Inter_Audit_Fee Mana_fee AuditFee ICQ DCE

TREAT −0.325 0.112*** 0.025 2.029 −0.000 (−1.18) (2.65) (0.57) (0.52) (−1.08)

AFTER −0.433* −0.028 −0.029 8.499 −0.000* (−1.71) (−0.67) (−0.76) (1.02) (−1.76)

TREAT � AFTER 0.584** 0.064** 0.222*** −20.048*** −0.000** (2.39) (2.16) (8.17) (−4.07) (−2.32)

Control variables Yes Yes Yes Yes Yes YEAR Yes Yes Yes Yes Yes INDUSTRY Yes Yes Yes Yes Yes N 2195 6382 6699 5438 6981 R2 0.492 0.776 0.625 0.365 0.286

Panel B: The Influence of Cost Pressure on Earnings Management

Variables

(1) (2) (3) (4)

Low financial performance

pressure

High financial performance

pressure

Low financial performance

pressure

High financial performance pressure

DA DA REM ROA REM ROA

TREAT −0.020*** −0.026*** −0.041 −0.044** (−2.62) (−3.08) (−1.59) (−2.40)

AFTER 0.001 −0.008 0.114*** −0.001 (0.14) (−0.67) (4.05) (−0.06)

TREAT � AFTER 0.009 0.026*** −0.008 0.061*** (1.10) (2.73) (−0.40) (3.29)

Control variables Yes Yes Yes Yes YEAR Yes Yes Yes Yes INDUSTRY Yes Yes Yes Yes Coefficient difference 1.97 6.50**

(0.160) (0.011) N 3303 3674 3143 3611 R2 0.064 0.128 0.189 0.194

***, ** and * indicate statistical significance at 1%, 5%, and 10% levels respectively (two-tailed).

EMERGING MARKETS FINANCE AND TRADE 3449

companies; Mana_fee represents the natural logarithm of the administrative fees of the listed compa- nies for the current year; Audit_fee denotes the natural logarithm of the listed firms’ annual report audit fees. The results are displayed in Columns (1)-(3) of Panel A of Table 6. The interaction term TREAT � AFTER is significantly and positively correlated with all the three measures of institutional compliance costs, indicating that the implementation of a mandatory internal control system caused the institutional compliance costs to increase.

To test the second dimension of the channel, we conduct a group test based on the firms’ financial performance pressure, proxied by an indicator of whether the corporate financial performance was higher or lower than the average industrial level. We expect that higher financial performance pressure would create stronger incentives for managers to manipulate the corporate financial performance, leading to a stronger effect of the “Announcement” on the earnings management. The results are reported in Panel B of Table 6. The coefficients of TREAT � AFTER in the group with low perfor- mance pressure is insignificant, while those in the high-pressure group are significantly positive. This finding holds for both the accrual-based and the real earnings management. These results indicate that the stimulating effect of the “Announcement” on the firms’ earnings management is stronger in the group with higher performance pressure, while no such an effect exists in the group with lower pressure. Taken together, the channel is verified.

5.2. More Effects of the Mandatory Internal Control in a Transitional Economy

As discussed above, the extra institutional compliance costs may create resistance in managers and employees toward the implementation of an internal control system. To verify this logic, we replace the dependent variable of Model (8) with internal control quality (ICQ), proxied by the Dibo Internal Control Index of the Chinese Listed Companies. The Dibo Index consists of comprehensive indicators of internal control qualities such as the implementation of corporate strategy, the realization of business objectives, and asset security. It covers the five internal control objectives stipulated in the “Standards,” and thus is able to fully measure firms’ internal control quality. In line with Schroeder and Shepardson (2015), we add two more control variables5 to Model (8). The results are reported in Column (4) of Panel A of Table 6. The interaction term TREAT � AFTER is significantly and negatively correlated with ICQ, indicating that the implementation of a mandatory internal control system had diminished the internal control quality. This means that the internal control quality depends on both the design and the contents of the system as well as the implementation mode.

To further confirm that the increase in earning management was driven by the firms’ cost pressure, we consider another type of earnings management, i.e., classification shifting. In contrast to the traditional accrual-based or real earnings management, classification shifting only changes the earn- ings structure within the income statement, without affecting the firms’ net profit or actual operating performance (McVay 2006). Based on the logic of this study, high institutional compliance costs create strong incentives for enterprises to manipulate profits upward, which means that the impact of a mandatory internal control on earnings management would be more significant. Since classification shifting does not affect the book value of the firms’ profit and thus does not help to alleviate the cost pressure, we expect that the management would reduce the use of classification shifting after the implementation of a mandatory internal control system. To test this conjecture, we replace the dependent variable of Model (8) with classification shifting (DCE).6 The results are displayed in Column (5) of Panel A of Table 6. The interaction term TREAT � AFTER is significantly and negatively correlated with DCE, indicating that the implementation of a mandatory internal control system would influence firms’ preference for different types of earnings management. To be more exact, the management prefers accrual-based and real earnings management to classification shifting. This denotes that the purpose of managers’ using real and accrual-based earnings management is to both meet and beat their financial performance targets, and to compensate for the extra cost pressure caused by the mandatory internal control.

3450 Z. SONG ET AL.

5.3. Long-term Effect of Mandatory Internal Control on Earnings Quality

Given that what we have found about the effect of the mandatory implementation of internal control on earnings quality may be short-term effects, the mandatory implementation of internal control may not affect the earnings quality in the long run. To test this conjecture, we delete observations with AFTER equal to 1 in the whole sample, and add observations of 2015–2016, to examine the long-term effect of mandatory implementation of internal control on accrual-based and real earnings manage- ment of listed companies. Table 7 presents the results. We find that, there is no significant difference in earnings management between listed companies subject to mandatory internal control and those otherwise. That is, although the mandatory implementation of internal control draws a negative impact on firms’ earnings quality in the short term, this impact will disappear in the long run.

6. Conclusion

Taking the promulgation of the “Announcement” by the MoF of China and the CSRC as the exogenous event. We find that, the mandatory implementation of an internal control system in China is more likely to stimulate the firms’ real and accrual-based earnings management. Furthermore, we find that the extra institutional compliance cost is an important channel through which mandatory internal control stimulates the firms’ earnings management. In addition, the results show that the firms’ internal control quality did not improve after the promulgation of the “Announcement,” which may be the result of the resistance brought by the extra compliance cost among the firms’ management and employees. In examining the impact of mandatory internal control on classification shifting, we find that, unlike accrual-based and real earnings management, classification shifting was reduced significantly after the “Announcement.” This finding indicates that the increase in the firms’ earnings management was driven by the managers’ incentive to beat or meet their financial performance targets, and thus confirms our conclusion that the extra cost pressure imposed by the mandatory internal control was the main cause of the increase in the firms’ earnings management.

These findings have several implications both for policy-makers and for the research related to internal control practices. First, in contrast to that on the developed markets, such as the US, where the implementation of the SOX Act has improved firms’ internal control quality, mandatory internal control in emerging economies, such as China, would have contradictory effects, causing the dete- rioration of internal control quality, and even more, a decline in earnings quality. Therefore, manda- tory internal-control-related studies should differentiate the research contexts to provide more reliable conclusions. In this sense, this study provides new theoretical and empirical evidence for the reform of the internal control systems in transitional economies. Second, there are practical implications in the finding that the excessive institutional compliance cost brought by the mandatory internal control was the main cause of the increase in earnings management. Thus, improving measures could be taken on

Table 7. Long-term effect of mandatory internal control on earnings quality.

(1) (2) (3) (4) (5)

DA REM_ROA AB_PROD_ROA AB_DISX_ROA AB_OCF_ROA

TREAT −0.008 −0.044** −0.043*** 0.001 0.000 (−1.48) (−2.33) (−3.29) (0.23) (0.05)

AFTER −0.030*** −0.051** −0.153*** −0.045*** −0.056*** (−4.01) (−2.02) (−7.49) (−7.50) (−7.19)

TREAT×AFTER 0.009 0.030 0.030 0.004 −0.004 (1.43) (1.26) (1.47) (0.88) (−0.55)

Control variables Yes Yes Yes Yes Yes YEAR Yes Yes Yes Yes Yes INDUSTRY Yes Yes Yes Yes Yes N 5818 5789 5789 5789 5789 R2 0.177 0.172 0.173 0.202 0.200

***, ** and * indicate statistical significance at 1%, 5%, and 10% levels respectively (two-tailed).

EMERGING MARKETS FINANCE AND TRADE 3451

both regulatory and implementing levels. On the regulatory level, the internal control system should be further improved, and the responsibilities for internal control information disclosure and internal control violations should be further clarified. On the implementation level, enterprises should link internal control responsibility with the employees’ performance evaluation, which would help to reduce the institutional compliance costs and diminish the incentives for earnings manipulation.

Notes

1. The “Announcement” stipulates that both the central and the local state-owned companies listed on the Main Board should fully implement an internal control system beginning in 2012; the non-state-owned companies listed on the Main Board with a total market capitalization of more than RMB 5 billion before December 31, 2011, and with an average net profit of more than RMB 30 million between 2009 and 2011 should fully implement an internal control system beginning in 2013; Other companies listed on the Main Board should fully implement an internal control system beginning in 2014.

2. Date source: FEI Survey: Sarbanes-Oxley Compliance Costs Are Dropping[R]. Financial Executives Research Foundation, 2006, http://www.fei.mediaroom.com/index.php?s=43&item=74.

3. Date source: SEC. Final Rule: Management’s Reports on Internal Control over Financial Reporting and Certification of Disclosure in Exchange Act Periodic Reports (Release No. 33–8238) [S]. Washington, D. C.: SEC Government Printing Office, 2003, http://www. Sec.gov/rules/final/32-8238.htm.

4. This is formulated in the “Standards.” 5. They are a dummy of whether there is a material weakness in the internal control (MW), and the natural

logarithm of the sum of the inventory turnover length and the accounts receivable turnover length (OP CYCLE). 6. For the calculation of classification shifting please refer to McVay (2006).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Funding

Jiaxin Wang acknowledges the financial support from National Natural Science Foundation of China [Project No. 72102229], Ministry of Education in China [Project No. 20YJC630143] and Zhongnan University of Economics and Law [Project No. 2722021BZ030]. Zilong Song acknowledges the financial support from Ministry of Education in China [Project No. 19YJC790113], National Natural Science Foundation of China [Project No. 71802105].

ORCID

Jiaxin Wang http://orcid.org/0000-0003-2110-5111

References

Bartov, E., D. Givoly, and C. Hayn. 2002. The rewards to meeting or beating earnings expectations. Journal of Accounting and Economics 33:173–204. doi:10.1016/S0165-4101(02)00045-9.

Chung, R., M. Firth, and J.-B. Kim. 2002. Institutional monitoring and opportunistic earnings management. Journal of Corporate Finance 8:29–48. doi:10.1016/S0929-1199(01)00039-6.

Cohen, D. A., A. Dey, and T. Z. Lys. 2008. Real and accrual-based earnings management in the pre-and post-Sarbanes- Oxley periods. The Accounting Review 83:757–87. doi:10.2308/accr.2008.83.3.757.

Dechow, P. M., R. G. Sloan, and A. P. Sweeney. 1995. Detecting earnings management. The Accounting Review 70:193–225.

Gao, H., Z. Shen, Y. Li, X. Mao, and Y. Shi. 2020. Institutional investors, real earnings management and cost of equity: Evidence from listed high-tech firms in China. Emerging Markets Finance and Trade 56:3490–506. doi:10.1080/ 1540496X.2019.1650348.

Gong, Q., O. Z. Li, Y. Lin, and L. Wu. 2016. On the benefits of audit market consolidation: evidence from merged audit firms. The Accounting Review 91:463–88. doi:10.2308/accr-51236.

3452 Z. SONG ET AL.

Gunny, K. A. 2010. The relation between earnings management using real activities manipulation and future perfor- mance: Evidence from meeting earnings benchmarks. Contemporary Accounting Research 27:855–88. doi:10.1111/ j.1911-3846.2010.01029.x.

Iliev, P. 2010. The effect of SOX section 404: Costs, earnings quality, and stock prices. The Journal of Finance 65:1163–96. doi:10.1111/j.1540-6261.2010.01564.x.

Kinney, W. R., Jr, and M. L. Shepardson. 2011. Do control effectiveness disclosures require SOX 404(b) internal control audits? A natural experiment with small U.S. public companies. Journal of Accounting Research 49:413–48. doi:10.1111/j.1475-679X.2011.00400.x.

Lin, J. Y. 1989. An economic theory of institutional change: Induced and imposed change. Cato Journal. 9:1–33. Maher, M. W., and D. Weiss. 2010. Costs of complying with SOX - measurement, variation and investors’ anticipation.

SSRN working paper. McVay, S. E. 2006. Earnings management using classification shifting: An examination of core earnings and special

items. The Accounting Review 81:501–31. doi:10.2308/accr.2006.81.3.501. Qin, Y., J. Cai, and S. Wei. 2021. Earnings management when firms face mandatory contributions. China Finance Review

International 11:522–51. doi:10.1108/CFRI-01-2021-0020. Schroeder, J. H., and M. L. Shepardson. 2015. Do SOX 404 control audits and management assessments improve overall

internal control system quality? The Accounting Review 91:1513–41. doi:10.2308/accr-51360. Shen, H., M. Zhang, M. Wang, J. Zhang, and Z. Guo. 2020. Does teacher-turned businessmen curb accrual-based earning

management and real activity manipulation. Emerging Markets Finance and Trade 58: 655–667. Tan, Y., X. Tian, X. Zhang, and H. Zhao. 2015. The real effects of privatization: evidence from china’s split share

structure reform. SSRN working paper. Xu, Z., and P. Rao. 2021. Does guarantee network reduce firms’ earnings management? Emerging Markets Finance and

Trade 57:2604–15. doi:10.1080/1540496X.2021.1939669. Zang, A. Y. 2012. Evidence on the trade-off between real activities manipulation and accrual-based earnings

management. The Accounting Review 87:675–703. doi:10.2308/accr-10196. Zhu, T., M. Lu, Y. Shan, and Y. Zhang. 2015. Accrual-based and real activity earnings management at the back door:

Evidence from Chinese reverse mergers. Pacific-Basin Finance Journal 35:317–39. doi:10.1016/j.pacfin.2015.01.008.

EMERGING MARKETS FINANCE AND TRADE 3453

Copyright of Emerging Markets Finance & Trade is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.