External auditor types and the cost stickiness of listed companies

Shangkun Liang; Dong Chen; Xiaoli Hu

doi:10.1080/21697213.2014.982004

External auditor types and the cost stickiness of listed companies

Liang, Shangkun; Chen, Dong; Hu, Xiaoli 2014-10-02 00:00:00 China Journal of Accounting Studies, 2014 Vol. 2, No. 4, 294–322, http://dx.doi.org/10.1080/21697213.2014.982004 a b c Shangkun Liang *, Dong Chen and Xiaoli Hu China’s Management Accounting Research & Development Center/School of Accountancy, Central University of Finance and Economics, Beijing 100081, China; School of Economics and Management, Wuhan University, Wuhan, 430000, China; School of Business, National University of Singapore, Kent 119077, Singapore External auditing is an important method of corporate governance that can effec- tively monitor and restrict a manager’s opportunistic behaviour and reduce cost stickiness. We focus on Chinese A-shares listed companies from 2002 to 2010 to explore the governance role of external auditors in cost stickiness. We classify audit ﬁrms into three types: the Big Four International, the Big Ten Domestic, and the non-Big Ten Domestic. The results show that the cost stickiness of ﬁrms audited by the Big Four International is signiﬁcantly lower, while the cost stickiness difference between ﬁrms audited by the latter two types is not signiﬁcant. Further tests show that the restriction on cost stickiness for the Big Four International is more signiﬁ- cant when the largest shareholder’s shareholding ratio is lower or when the local market is less developed. These results indicate that the restriction placed by audit quality on cost stickiness varies with other corporate governance mechanisms. Our results are qualitatively unchanged under different robustness tests, including pro- pensity score matching. This study shows that Big Four International audit ﬁrms can better reduce cost stickiness, which provides evidence of the higher audit quality of these ﬁrms and extends the study of restrictive factors on cost stickiness to exter- nal corporate governance mechanisms. Keywords: Big Four International; cost stickiness; shareholding ratio; marketisation level; propensity score matching 1. Introduction External auditing is an indispensable part of internal corporate governance (Cohen, Krishnamoorthy, & Wright, 2002; Fan & Wong, 2005). The modern enterprise system features the separation of ownership and management. This mechanism might improve the efﬁciency of capital and human talents. However, it also inevitably leads to conﬂict between principals and agents (Jensen & Meckling, 1976). It is possible that agents (managers) with information superiority would harm the interest of principals (share- holders) to maximise their own interests. External audits can mitigate information asymmetry between contracting parties and reduce internal agency cost through inde- pendent audit work (Watts & Zimmerman, 1983). Many Chinese studies focus on the choice and impact of external auditors, including constraints imposed by external audi- tors on earnings management and ﬁnancial restatement (Cai, Huang, & Zhao, 2005; Li, Wang, & Wang, 2004; Liu, 2009; Liu & Yan, 2006; Wang & Zhang, 2005), and on related-party transactions (Gao & Wu, 2007; Hong, Xu, & Li, 2011; Yang, Lin, & *Corresponding author. Email: jevonacc@gmail.com Paper accepted by Xi Wu. © 2014 Accounting Society of China China Journal of Accounting Studies 295 Wang, 2009; Yue, 2006; Zeng & Ye, 2005). There is also a constraint through the governance role of the external auditors in the turnover of chief executive ofﬁcers (CEOs) (Jiang, Liu, & Liu, 2007; Wang, 2011). However, there are fewer studies on the constraints imposed by external auditing on managerial behaviour (Jiang et al., 2007) in this strand of literature. Such studies as have been carried out focus mainly on the quality of information disclosure, which is the ﬁnal output of the information dis- closure process. This study argues that external audit governance is effective not only in the ﬁnal part of information disclosure, but also during operational and production procedures. To explore this possibility, we examine the inﬂuence of external audit on cost stickiness. Cost stickiness refers to the phenomenon where costs decrease less when business activity falls than they increase when business activity rises (Anderson, Banker, & Janakiraman, 2003). Anderson et al. (2003) tested this phenomenon for the ﬁrst time in the ﬁeld of ﬁnance and accounting, using US data. We look at the association between external audit and cost stickiness in China for the following three reasons. First, this study allows us to explore the mechanism of for- mation of cost stickiness from an external perspective. The literature on cost stickiness is mainly based on an internal perspective. Anderson et al. (2003), Calleja, Steliaros, and Thomas (2006), Chen, Lu, and Sougiannis (2012), and Sun and Liu (2004) point out that managerial opportunistic behaviour is one of the primary causes of cost sticki- ness and they mainly focus on internal factors such as the structure of the board of directors. However, our study extends the investigation of the causes of cost stickiness to external governance mechanisms and looks at the role of external auditors for the ﬁrst time, which contributes to the current literature. Second, it explores and shows the incentive for the demand for high-quality external audits, given China’s institutional background. Ball, Robin, and Wu (2003) argue that the large number of related-party transactions in East Asian countries leads to a lack of motivation for listed ﬁrms to pro- vide high-quality accounting information. Defond, Wong, and Li (2000) also point out that, under government regulation, Chinese listed companies lack demand for high- quality audit services. However, the environment of Chinese listed ﬁrms has changed with the launch of non-tradable share reform, the development of marketisation, and the move to public listing of many non-state-owned ﬁrms. Since the conﬂict between shareholders and managers is becoming increasingly severe, the need to monitor man- agers can become one source of demand for high-quality audits. Third, it provides evi- dence for the debate on the audit quality of the Big Four International ﬁrms. Most international studies agree that the audit services of the Big Four International ﬁrms are of higher quality (Becker, Defond, Jiambalvo, & Subramanyam, 1998; Francis, Maydew, & Sparks, 1999; Lennox, 1999). However, although the same opinion is dominant in Chinese studies, controversy still exists (Guo, 2011; Liu & Xu, 2002; Liu & Zhou, 2007; Qi, Chen, & Zhang, 2004; Wang & Wang, 2006;Yu, 2007). We focus on the governance role of the Big Four International ﬁrms and the Big Ten Domestic audit ﬁrms in China in order to provide evidence for this debate from a new perspective. When there is a business boom, optimistic managers are likely to increase expenses (especially unproductive expenditures) drastically to enhance personal utility or to con- trol more resources (Chen, Lu, & Sougiannis, 2012). When there is a business reces- sion, however, managers may not be willing to decrease such expenses or may just decrease them slightly. Such behaviour results in high cost stickiness; that is, expenses decrease more slowly when revenue falls than expenses increase when revenue rises by the same amount. To the extent that high-quality external audits can effectively regulate 296 Liang et al. the corporate ﬁnancial process and constrain managers’ opportunistic behaviour (Watts & Zimmerman, 1983), we suggest high-quality external audits may help to reduce cost stickiness. Using Chinese A-shares listed companies from 2002 to 2010, we test the relationship between high quality external audit and cost stickiness. We classify audit ﬁrms into three types: the Big Four International, the Big Ten Domestic, and the non- Big Ten Domestic. The results show that the cost stickiness is signiﬁcantly lower in listed companies audited by the Big Four International, while there is no signiﬁcant cost stickiness difference between ﬁrms audited by the latter two types. In addition, the restriction on cost stickiness associated with the Big Four International is much more signiﬁcant when the largest shareholder’s shareholding ratio is lower or when the local market is less developed, which indicates that the audit quality’s restriction on cost stickiness varies with other corporate governance mechanisms. This study reveals that the Big Four International are associated with cost stickiness and extends the study on cost stickiness factors to external governance mechanisms. This paper contributes to the literature in the following three ways. First, it enriches the literature that explores factors associated with cost stickiness and sheds light on how external audit quality interacts with factors identiﬁed in prior literature. This paper investigates empirically for the ﬁrst time how auditor supervision, an external gover- nance method, impacts cost stickiness and ﬁnds that the Big Four International consid- erably inhibit cost stickiness. In addition, we ﬁnd that the impact of audit quality on cost stickiness varies with changes in other corporate governance mechanisms, such as ownership structure and external environment, which helps to understand cost stickiness factors and how these other factors affect cost stickiness. Second, it shows motivation for the demand for high-quality auditing in China. We ﬁnd the Big Four International ﬁrms are associated signiﬁcantly with cost stickiness (e.g. when the shareholding ratio of major shareholders is lower). Such a restrictive effect on cost stickiness could effec- tively increase the beneﬁts of hiring audit ﬁrms with higher audit quality and add to the demand for high-quality audit services. Third, we provide evidence of the quality differences between different audit ﬁrms from a new perspective. We show that while the Big Four International are signiﬁcantly associated with cost stickiness, the Big Ten Domestic ﬁrms do not have a similar impact. Our results are consistent with the argu- ment that the Big Four International produce higher-quality audit work. The rest of the paper is structured as follows. Section 2 presents the institutional background, literature review and hypothesis development. Section 3 describes the research design. Section 4 shows the empirical test results. Section 5 concludes and dis- cusses potential limitations. 2. Institutional background, literature review, and hypothesis development 2.1. Institutional background The development of external auditing in China is closely related to politics and the economy. After China’s reform and opening up (after 1978), the CPA (Certiﬁed Public Accountants) system was restored and the audit business redeveloped (Liu & Xu, 2002). According to the Temporary Regulations on the Examination and Approval Pro- cedures of Limited Liability Accounting Firms passed by the Ministry of Finance in 1993, early accounting ﬁrms were mostly established relying on government agencies, such as the local Finance Departments of the different provinces. This situation remained the same until April 1998, when the Ministry of Finance released The Notice about Decoupling Accounting Firms Which Have Securities or Futures Related China Journal of Accounting Studies 297 Business from Their Linked Units, which launched the decoupling reform. By the end of 1999, the decoupling reform was mostly accomplished. However, domestic account- ing ﬁrms that originally relied on government agencies still had many inherent deﬁcien- cies. Along with several serious incidents of fraud in China’s capital market, market attention to audit quality gradually grew, initiating the demand for the audit work of the original Big Five International ﬁrms. In 2001, the China Securities Regulatory Commission together with the Ministry of Finance released The Regulation on the Tem- porary License for Foreign Accounting Firms to Conduct Business in Financial Listed Firms. In the same year, the two regulatory departments authorised temporary licences to the Big Four (originally the Big Five) international accounting ﬁrms and required that listed ﬁrms in the banking, security, and insurance industry be audited by both domestic and international ﬁrms. Similar administrative regulations requiring certain types of ﬁrms to be audited by international audit ﬁrms were also implemented after- wards (Liu & Xu, 2002; Liu & Zhou, 2007). In 2001 China signed an agreement to open up the CPA service market when it joined the World Trade Organization (Qi et al., 2004), which led to the complete opening of the auditing market to the Big Four International. Over the past ten years, the Big Four International ﬁrms have developed rapidly in the audit market in China. Their total income accounted for more than 55% of the total income of China’s top 100 accounting ﬁrms at the end of 2007. After 2007, as a large number of small companies went public in the SME (small and medium sized enterprises) and GEM (growth enterprises market) boards, the income share of the Big Four International declined but still surpassed 30% of the total income of the top 100 accounting ﬁrms (according to data from The Comprehensive Evaluation of Top 100 Accounting Firms in China published by the Chinese Association of Certiﬁed Public Accountants from 2003 to 2013). In addition, the Big Four International audit ﬁrms are widely recognised as having professional advantages in human capital, accu- mulative knowledge, and non-audit services (Qi et al., 2004; Wang & Wang, 2006). 2.2. Literature review Anderson et al. (2003) were the ﬁrst in the ﬁeld of ﬁnance and accounting to propose the concept of cost stickiness, which refers to the phenomenon where costs decrease less when business activity falls than they increase when business activity rises. The authors use a sample of US-listed companies to test this theory empirically. They ﬁnd that selling, general, and administrative (SG&A) costs increase by 0.55% when sales revenue increases by 1% but only decrease by 0.35% when sales revenue decreases by 1%, which provides preliminary evidence of cost stickiness. Numerous studies further conﬁrm the existence of cost stickiness (Calleja et al., 2006; Chen, Lu, & Sougiannia, 2012; Dierynck, Landsman, & Renders, 2012) and they focus more on the causes of cost stickiness. Among these causes, managers’ opportu- nistic behaviour is considered to be crucial (Banker, Byzalov, & Plehn-Dujowich, 2011; Sun & Liu, 2004) and empirical evidence is relatively abundant. Calleja et al. (2006) compare the degree of cost stickiness in four countries. They point out that weaker cor- porate governance accounts for the higher cost stickiness in France and Germany than in the United States and the United Kingdom. Chen, Lu, and Sougiannis (2012) further examine the effect of agency costs between stockholders and managers on cost sticki- ness. They ﬁnd that cost stickiness is highly positively related to the incentive for man- agerial empire building (such as free cash ﬂow, executive tenure, expectation to leave, and compensation structure). They also ﬁnd that strong corporate governance can 298 Liang et al. mitigate this positive association. Dierynck et al. (2012) provide evidence that executives’ incentive to conduct earnings management affects labour cost stickiness. They ﬁnd that, to meet the goal of proﬁt, management tends to manipulate labour costs. Apart from managers’ opportunistic behaviour, adjustment cost and management opti- mistic expectations can also cause cost stickiness. Regarding adjustment cost, Anderson et al. (2003) propose that ﬁrms with large proportions of physical assets and human capital tend to have greater cost stickiness, since the adjustment costs of these two assets are higher. Banker, Byzalor, and Chen (2013) ﬁnd that the power of a country’s labour union has a signiﬁcant inﬂuence on the adjustment costs of human resource. If a county’s labour union is more powerful, the adjustment cost of human resources will be greater and labour cost stickiness will be correspondingly higher. Regarding man- agement optimistic expectations, Anderson et al. (2003) suggest that when the macro- economy grows rapidly, management will be optimistic about the future, even after a short-term decline in revenue, so cost stickiness will be greater. Similarly, Banker et al. (2011) ﬁnd that companies in growing industries have greater cost stickiness. In terms of Chinese domestic studies, Sun and Liu (2004) ﬁnd that cost stickiness also exists in China’s listed companies, but macroeconomic growth and capital inten- sive cannot explain it well. Therefore, managers’ opportunistic behaviour may be a bet- ter explanation. Wan and Wang (2011) ﬁnd that CEO non-duality and a higher proportion of independent directors can effectively reduce cost stickiness. Chen, Song, and Shi (2012) propose that controlling for managers’ incentives to manage earnings, as in a ‘big bath’, can drastically decrease the estimated cost stickiness. Liang (2013) ﬁnds that cost stickiness is negatively associated with the level of ownership concentration. Given the ﬁndings of the foreign and domestic literature, there have been many studies on the impact of managers’ opportunistic behaviour on cost stickiness. How- ever, these studies mainly focus on the effect of internal corporate governance mecha- nisms rather than external governance mechanisms. Based on the studies above, we further include audit supervision, an important external governance factor, in this line of study and examine the effect of external audit quality on the cost stickiness of listed companies. 2.3. Hypothesis development In this paper, we seek to develop the argument that that high-quality external audits can reduce listed companies’ cost stickiness. Managers’ opportunistic behaviour is a main cause of cost stickiness (Banker et al., 2011). In modern corporations, with the separation of management and ownership, minority shareholders have to pay large amounts to obtain comprehensive information about companies’ operation and develop- ment, while managers do not (Jensen & Meckling, 1976). In addition, since executive incentive contracts cannot completely reconcile shareholders’ and managers’ interests, managers pervasively have an ‘empire building’ incentive. Managers tend to over- expend ﬁrm size and spend a great amount of resources on personal gains instead of improving company value (Jensen, 1986; Stulz, 1990). When business performance declines, managers’ optimal behaviour should be to cut down costs and keep them in accordance with activities. However, managers with strong opportunistic incentives will not adjust or only slowly adjust costs when sales decline (Chen, Lu, & Sougiannis, 2012). For example, to pursue personal prestige and rights, managers hire large num- bers of employees, frequently update ofﬁce equipment, and incur many other expenses. China Journal of Accounting Studies 299 In such a situation, costs decrease less when business activity falls than costs increase when business activity rises and we regard this asymmetric cost change as cost stickiness. We propose that high-quality external audits can decrease managers’ opportunistic behaviour and further help decrease cost stickiness. Becker et al. (1998) and Francis et al. (1999) point out that high-quality external audits have a great inﬂuence on corpo- rate behaviour, such as standardising the ﬁnancial procedures of listed companies, improving information transparency, and reducing ﬁnancial fraud and earnings manage- ment. High-quality external audits can identify and correct management fraud and earnings management (Qi et al., 2004). They can also alert shareholders by providing non-standard audit opinions (Jiang et al., 2007), which indirectly limits managers’ opportunistic behaviour. Under weak audit supervision, managers can package costs or conduct earnings management to hide the negative impact of such costs on company performance. However, under strong audit supervision, it is difﬁcult to hide the nega- tive impact. Considering the difﬁculties of conducting ﬁnancial fraud or earnings man- agement, managers will decrease their opportunistic behaviour under strong audit supervision and cost stickiness will decrease. In addition, through standardising ﬁnan- cial procedures, high-level external audits can also improve the quality of ﬁnancial report information to help shareholders compare the information with other companies in the same industry accurately and in a timely manner and identify the relation of abnormal revenues and abnormal costs. Shareholders can thus make better decisions about executive compensation and replacement and strengthen control on management. Previous studies show that Big Four International ﬁrms, with high-level specialisa- tion and independence, can represent high-quality external auditing. Dopuch and Simu- nic (1980) ﬁnd that big accounting ﬁrms tend to undergo more specialised training, stricter certiﬁcation, and peer review. O’Keefe and Westort (1992) ﬁnd that big accounting ﬁrms invest more in knowledge. When the size of auditor grows, the quasi- rent for each client will decrease. Then the probability of an audit ﬁrm behaving oppor- tunistically will decrease, so the audit ﬁrm is more independent (DeAngelo, 1981). Dye (1993), Moore and Scott (1989) and Watts and Zimmerman (1981, 1986) demonstrate that big accounting ﬁrms are more independent from the perspective of the brand repu- tation mechanism and deep pocket theory. Greater ability helps audit ﬁrms identify and correct ﬁnancial procedure weakness and restrict managers’ earnings management. If the accounting ﬁrms are more independent, they are more likely to provide objective and fair audit reports and can increase deterrence to managers. Compared with domes- tic audit ﬁrms, Big Four International ﬁrms have obvious advantages due to their his- tory, size, specialisation, market share, and so forth, which ensure their higher audit quality (Qi et al., 2004). In addition, Wang, Wong, and Xia (2008) ﬁnd that, in China, government intervention can reduce audit quality, especially the quality of small local audit ﬁrms. We can infer from the previous analysis that, compared with domestic ﬁrms, the Big Four International provide better audit services and companies audited by them tend to have lower degrees of cost stickiness. Similarly, according to the quasi-rent and deep pocket theories, the Big Ten Domestic audit ﬁrms should have higher audit quality than other domestic ﬁrms and their clients’ cost stickiness should therefore be lower than that of the clients of other domestic ﬁrms. The above discus- sion leads to our ﬁrst hypothesis. H1a: Compared with listed companies audited by domestic audit ﬁrms, companies audited by the Big Four International have a lower level of cost stickiness. 300 Liang et al. H1b: Compared with listed companies audited by non-Big Ten Domestic audit ﬁrms, companies audited by the Big Ten Domestic have a lower level of cost stickiness. Our prior analysis suggests high-quality external audit can mitigate cost stickiness by reducing agency costs between stockholders and managers. However, external audit is only part of corporate governance and other governance mechanisms can also reduce agency costs and control managers’ opportunistic behaviour. If other mechanisms can successfully reduce agency costs, external audit may not be the dominant factor in the supervision, but if other mechanisms fail, external audit can be more important. Speciﬁcally, we analyse the interaction between different forms of external gover- nance regarding two aspects: ownership structure and the external environment. The largest shareholder’s holding ratio and marketisation level of the province where a ﬁrm’s headquarter is located represent two important dimensions of external governance mechanisms and are relatively exogenous to the Big Four International indicator. Since these two factors can vary cross-sectionally, the degree of corporate agency prob- lem can vary accordingly. If the largest shareholder possesses a higher proportion of shares, it will monitor the managers more actively (Shleifer & Vishny, 1986). The pro- portion of shares held by the largest shareholder is negatively associated with agency costs. Chen and Liang (2010) ﬁnd that non-pecuniary compensation is attenuated when the largest shareholder holds a higher proportion of shares. Dou, Liu, and An (2011) ﬁnd that large shareholders’ complete control of a ﬁrm can prevent inefﬁcient invest- ments. In addition, when the market is highly developed, listed companies will face more competition and less government intervention and investors can identify manag- ers’ efforts accurately and in a timely fashion. As a result, a high level of marketisation leads to low agency costs. Li (2007) ﬁnds that ﬁrms’ SG&A rate and efﬁciency loss both decrease when market competition increases. Jiang, Huang, and Zhang (2009) ﬁnd that administrative expense rates decrease and asset turnover increases in highly com- petitive product markets. When the largest shareholder holds a lower proportion of shares and the market level is lower, agency costs are relatively higher and an external audit can play a more important role in reducing cost stickiness. The above discussion leads to our second hypothesis. H2a: When the largest shareholder holds a lower proportion of shares, the negative associ- ation between being audited by the Big Four International or Big Ten Domestic and cost stickiness is more pronounced. H2b: When the market level is lower, the negative association between being audited by the Big Four International or Big Ten Domestic and cost stickiness is more pronounced. 3. Research design 3.1. Sample selection Our initial sample starts with all A-shares listed ﬁrms in the Shanghai and Shenzhen Stock Exchanges from 2002 to 2010. We then delete 1670 observations that were listed for less than two years and 133 observations in the ﬁnancial industry. We also delete observations whose change in operating revenue or expense from the previous year is lower than 0.5% or higher than 99.5%, which includes 723 observations (Chen, Lu, & Sougiannis, 2012). Finally, we delete 534 observations with missing controlling variables. Our ﬁnal sample includes 10,494 observations. Table 1 shows the sample selection procedures. China Journal of Accounting Studies 301 Table 1. Sample selection process. Listed less Firms with extreme Initial than 2 Financial changes in income or Missing Remaining Year observations years companies expense (0.5%) data observations 2002 1,211 150 7 64 56 934 2003 1,278 138 7 71 43 1,019 2004 1,377 167 8 88 34 1,080 2005 1,366 115 10 88 55 1,098 2006 1,421 81 10 98 65 1,167 2007 1,537 192 14 107 52 1,172 2008 1,611 203 17 47 72 1,272 2009 1,703 176 30 80 82 1,335 2010 2,050 448 30 80 75 1,417 Total 13,554 1,670 133 723 534 10,494 3.2. Research design and variable deﬁnitions Following Anderson et al. (2003), Chen, Lu, and Sougiannis (2012) and Sun and Liu (2004), we establish the following model to test our hypothesis: DCost ¼ a þ a DIncome þ a DIncome Income decrease þ a DIncome 0 1 2 3 Income decrease Big4 þ a Big4 þ a DIncome Income decrease 4 5 Domestic big10 þ a Domestic big10 þ RControl þ e (1) where ΔCost and ΔIncome represent changes in expense and in operating income, respectively. We ﬁrst calculate the ratio of expenses in year t to expense in year t – 1 and the ratio of the operating income in year t to that in year t – 1 and then take the natural logarithm of these two ratios. Speciﬁcally, expense is the sum of operating cost and general cost; Income_decrease is a dummy variable for income decline, which equals one if the operating income in year t was lower than that in year t – 1 and zero otherwise; Big4 is an indicator variable for Big Four International audit ﬁrms that equals one if the listed company was audited by Ernst & Young, KPMG, Deloitte, or PwC in year t and zero otherwise; and Domestic_big10 is a dummy variable of Big Ten Domestic audit ﬁrms that equals 1 if the ﬁrm was audited by a Big Ten Domestic audit ﬁrm and zero otherwise. Following Anderson et al. (2003), Chen, Lu, and Sougiannis (2012) and Sun and Liu (2004), our control variables include an indicator variable for Income decline for two consecutive years (Two_year_decrease), economic growth (GDP_growth), human capital intensity (Employee_intensity), ﬁxed capital intensity (Fixed_asset_intensity), ﬁnancial leverage (Leverage), the market index (Marketisation_index), proportion of independent directors (Independent_director), CEO duality (Duality), and proportion of managerial shareholdings (Managerial_share). Table 2 shows the deﬁnitions and explanations of the main variables. To reduce the effect of extreme values, all continuous variables are winsorized at the 0.5% level. We control for year ﬁxed effects and clustered observations by ﬁrms (Petersen, 2009)to mitigate the effect of serial correlation. We also report robust t-values to adjust for het- eroscedasticity (White, 1980). These methods improve the robustness of our empirical results. If cost stickiness exists, the regression coefﬁcient a in model (1) should be signiﬁ- cantly negative. If H1a holds, we would observe that a in model (1) is signiﬁcantly 3 302 Liang et al. Table 2. Variable deﬁnitions. Type Variables name Variables symbol Variables deﬁnition Dependent Changes in the ΔCost Natural logarithm of the ratio of the variable costs expense in year t to the expense in year t–1 Independent Changes in ΔIncome Natural logarithm of the ratio of the variables operating income operating income in year t to the operating income in year t–1 Income declining Income_decrease Dummy variable that equals 1 if operating income in year t is lower than that in year t–1, and equals 0 otherwise The Big Four Big4 Dummy variable that equals 1 if the International ﬁrm is audited by Big Four International, 0 otherwise The Big Ten Domestic_big10 Dummy variable that equals 1 if the Domestic ﬁrm is audited by Big Ten Domestic, 0 otherwise Control Income declining Two_year_decrease Dummy variable that equals 1 if the variables for two operating income declines for two consecutive years consecutive years, and equals 0 otherwise Economic growth GDP_growth GDP growth for year t Human capital Employee_intensity Number of employees/operating revenue intensity (in millions) Fixed capital Fixed_asset_intensity Total assets/operating revenue intensity Financial leverage Leverage Debts/assets Marketisation Marketisation_index Marketisation index (Fan et al., 2010) index Proportion of Independent_director Proportion of independent directors independent directors CEO duality Duality Dummy variable that equals 1 if the CEO and Chairman are the same one, otherwise 0 Proportion of Managerial_share Proportion of shares held by managers management shareholdings positive. If H1b holds, a should be signiﬁcantly positive. If H2 holds, we expect that a and a in model (1) exhibit signiﬁcant differences between groups with different 3 5 ownership structures or local marketisation levels. 3.3. Descriptive statistics and correlation coefﬁcients Table 3 reports the descriptive statistics of the main variables. Panel A shows the descriptive statistics for the full sample. The means of ΔCost and ΔIncome are 0.133 and 0.145, respectively, and the ratio of observations with an income decrease (mean of Income_decrease) is 25.6%. These statistics are close to those reported by Gong, Liu, and Shen (2010). The proportion of ﬁrm-years audited by Big Four International (mean of Big4) is 5.65%, which is close to the result of Guo (2011), Wang and China Journal of Accounting Studies 303 Table 3. Descriptive statistics. Panel A: Full sample Variables symbol Mean Lower quartile Median Upper quartile Standard deviation ΔCost 0.133 –1.499 0.132 1.752 0.375 ΔIncome 0.145 –1.525 0.141 2.151 0.391 Income_decrease 0.256 0.000 0.000 1.000 0.436 Big4 0.057 0.000 0.000 1.000 0.231 Domestic_big10 0.335 0.000 0.000 1.000 0.472 Two_year_decrease 0.096 0.000 0.000 1.000 0.295 GDP_growth 10.726 9.100 10.100 14.200 1.618 Employee_intensity 2.717 0.023 1.700 28.418 3.503 Fixed_asset_intensity 2.717 0.304 1.794 36.022 3.630 Leverage 0.506 0.062 0.515 0.957 0.186 Marketisation_index 8.060 0.630 7.920 11.710 2.202 Independent_director 0.346 0.000 0.333 0.600 0.073 Duality 0.096 0.000 0.000 1.000 0.295 Managerial_share 0.005 0.000 0.000 0.273 0.028 Panel B: Sub-sample Big Four Big Four Big Ten Non-Big Ten International–Non- International Domestic Domestic Big Ten Domestic Variables symbol Mean Median Mean Median Mean Median Mean Median ΔCost 0.145 0.127 0.138 0.133 0.128 0.132 0.98 1.44 ΔIncome 0.165 0.154 0.145 0.144 0.145 0.138 0.76 0.08 Income_decrease 0.189 0.000 0.260 0.000 0.259 0.000 –3.84*** –3.83*** Two_year_decrease 0.057 0.000 0.093 0.000 0.101 0.000 –3.30*** –3.29*** GDP_growth 10.647 10.100 10.630 10.100 10.786 10.100 –1.22 –1.47 Employee_intensity 1.657 0.994 2.239 1.508 3.079 1.886 –7.60*** –10.87*** Fixed_asset_intensity 2.261 1.605 2.503 1.670 2.877 1.893 –3.15*** –4.37*** Leverage 0.471 0.478 0.517 0.531 0.503 0.512 –4.74*** –4.87*** Marketisation_index 8.826 8.960 8.580 8.660 7.703 7.420 8.74*** 8.88*** Independent_director 0.341 0.333 0.355 0.333 0.342 0.333 –1.88* –1.24 Duality 0.057 0.000 0.103 0.000 0.096 0.000 –3.29*** –3.29*** Managerial_share 0.001 0.000 0.006 0.000 0.004 0.000 –3.39*** –0.75 Note: For deﬁnitions of the variables, please refer to Table 2. ***, **, * indicate signiﬁcance levels of 1%, 5% and 10% respectively Wang (2006), Yu (2007), and others. A total of 33.5% of observations are audited by Big Ten Domestic (mean of Domestic_big10). All the statistics of the control vari- ables are within reasonable ranges. Panel B reports the descriptive statistics for subs- amples partitioned by the types of audit ﬁrms. The results show that the overall condition is better in listed companies audited by Big Four International ﬁrms. Finan- cial leverage (Leverage), of the likelihood of having a decreased income, the likeli- hood of two consecutive years of decreased income, and the likelihood of CEO duality are lower for such ﬁrms and, on average, the local marketisation level is higher. These results are consistent with previous studies (e.g. Qi et al., 2004; Wang & Wang, 2006). Table 4 shows the correlation coefﬁcients of the main variables. The correlation matrix shows no strong correlation between the independent variables and the control variables. 304 Liang et al. Table 4. Correlation coefﬁcients. AB C D EF G H I J K L M (A) ΔCost 1 *** (B) ΔIncome 0.294 1 *** *** (C) Income_decrease –0.200 –0.536 1 *** (D) Big4 0.007 0.010 –0.037 1 *** (E) Domestic_big10 0.009 –0.001 0.008 –0.174 1 *** *** *** *** (F) Two_year_decrease –0.142 –0.344 0.556 –0.032 –0.008 1 *** *** *** *** *** (G) GDP_growth –0.073 0.045 –0.088 –0.012 –0.042 –0.040 1 *** *** *** *** *** *** *** (H) Employee_intensity –0.042 –0.246 0.178 –0.074 –0.098 0.220 –0.057 1 *** *** *** ** *** *** *** *** (I) Fixed_asset_intensity –0.058 –0.353 0.244 –0.031 –0.042 0.267 –0.042 0.410 1 *** *** *** *** *** ** (J) Leverage 0.048 0.047 0.005 –0.046 0.043 –0.005 0.043 –0.026 –0.006 1 *** ** *** *** *** *** *** (K) Marketisation_index –0.047 –0.019 0.029 0.085 0.168 0.003 0.124 –0.261 –0.066 –0.012 1 ** *** * *** *** ** * *** (L) Independent_director –0.031 0.016 –0.005 –0.018 0.087 –0.025 0.117 –0.129 –0.027 0.021 0.223 1 ** ** *** * (M) Duality 0.005 –0.000 –0.009 –0.032 0.017 –0.005 0.001 0.011 0.007 –0.031 0.038 0.020 1 ** ** *** *** *** * *** *** *** *** *** (N) Managerial_share 0.026 0.012 –0.031 –0.033 0.036 –0.033 –0.015 –0.024 –0.034 –0.081 0.144 0.071 0.161 Note: For deﬁnitions of the variables, please refer to Table 2. ***, **, * indicate signiﬁcance levels of 1%, 5% and 10% respectively. China Journal of Accounting Studies 305 4. Empirical analysis 4.1. Testing H1: types of external auditors and cost stickiness Table 5 shows the regression results of model (1). Column (1) contains only ΔIncome and ΔIncome*Income_decrease, while columns (2) to (6) further include control vari- ables and our variables of interest. The regression result shows that the F-values of the regression model are mostly above 20 and the adjusted R values are about 0.10, close to the results of Sun and Liu (2004). The coefﬁcient of ΔIncome*Income_decrease is negative and signiﬁcant at the 1% level in column (1), suggesting that cost stickiness exists in China’s listed companies, which is consistent with the results of Gong et al. (2010) and Sun and Liu (2004). The sign and signiﬁcance of the coefﬁcient of ΔIncome*Income_decrease remain unchanged in column (2) when control variables are included. Columns (3) to (5) present the regression results for the subsamples by type of external auditors. The coefﬁcient of ΔIncome*Income_decrease is positive and not signiﬁcant in column (3) for companies audited by Big Four International ﬁrms, nega- tive but not signiﬁcant in column (4) for companies audited by Big Ten Domestic ﬁrms, and negative and signiﬁcant at the 5% level in column (5) for companies audited by other audit ﬁrms. Columns (6) and (7) include the interaction term of three variables in the model to test for H1a and H1b. The coefﬁcient of ΔIncome*Income_ decrease*Big4 is positive and signiﬁcant at the 10% level in column (6), meaning that Big Four International audited ﬁrms have signiﬁcantly less cost stickiness compared with other ﬁrms. Column (7) further includes the interaction of Domebig10 and ΔIncome*Income_decrease. The coefﬁcient of ΔIncome*Income_decrease*Big4 remains positive and signiﬁcant at the 10% level, but the coefﬁcient of ΔIncome*Inco- me_decrease*Big10 is positive but not signiﬁcant, which means that there is no signiﬁ- cant difference in cost stickiness between ﬁrms audited by Big Ten Domestic and by the non-Big Ten Domestic. The results in columns (3) to (7) support H1a but not H1b. We think that the non-signiﬁcant result for H1b is not surprising. For different reasons, such as a less diversiﬁed business, brain drain, and government intervention, most pre- vious studies have failed to ﬁnd signiﬁcant audit quality differences between domestic audit ﬁrms of different sizes (Fang, Hong, & Li, 2004;Wu, 2006; Yuan & Li, 2003). In addition (untabled), the joint test shows that the sum of the coefﬁcients of ΔIncome*Income_decrease and ΔIncome*Income_decrease*Big4 is not signiﬁcantly different from zero. 4.2. Testing H2: types of external auditors, agency costs, and cost stickiness For H2, we separate the entire sample into two subsamples according to the proportion of shares held by the largest shareholders. If the proportion is above the median, the observation is classiﬁed as a high stake of major shareholders; otherwise the observa- tion is classiﬁed as a low stake of major shareholders. We want to use subsample regressions to explore the difference in the restrictive roles of Big Four International and Big Ten Domestic audit ﬁrms on cost stickiness in different environments. Table 6 shows the regression results. The coefﬁcient of ΔIncome*Income_ decrease*Big4 is not signiﬁcant in column (1) for ﬁrms with high stakes of major shareholders, suggesting that Big Four International audit ﬁrms do not signiﬁcantly reduce cost stickiness when the largest shareholder’s holding ratio is high. However, the coefﬁcient of ΔIncome*Income_decrease*Big4 is positive and signiﬁcant at the 1% level in column (2) for ﬁrms with low stakes of major shareholders, which means that 306 Liang et al. Table 5. Types of external auditors and the cost stickiness. (1) Full (2) Full (3) The Big (4) The Big Ten (5) Others (6) Full (7) Full sample sample Four Domestic domestic sample sample 0.309*** 0.294*** 0.351*** 0.332*** 0.274*** 0.294*** 0.294*** (10.68) (10.10) (3.51) (6.69) (7.40) (10.10) (10.08) ΔIncome*Income_decrease –0.151*** –0.145** 0.076 –0.116 –0.153** –0.148*** –0.169*** (–3.01) (–2.57) (0.48) (–1.45) (–2.14) (–2.61) (–2.71) ΔIncome*Income_decrease*Big4 0.214* 0.238* (1.70) (1.84) 0.024** 0.031*** (2.20) (2.71) ΔIncome*Income_decrease 0.086 *Domestic_big10 (1.27) Domestic_big10 0.020*** (2.68) Two_year_decrease –0.085*** –0.041 –0.078*** –0.086*** –0.085*** –0.083*** (–4.86) (–0.57) (–3.08) (–3.77) (–4.83) (–4.82) GDP_growth –0.022*** 0.014 –0.022*** –0.025*** –0.022*** –0.022*** (–5.43) (1.26) (–2.65) (–4.95) (–5.35) (–5.40) Employee_intensity 0.001 –0.008 0.001 0.001 0.001 0.001 (0.39) (–1.50) (0.34) (0.28) (0.42) (0.38) Fixed_asset_intensity 0.002 0.017*** 0.003 0.002 0.002 0.002 (1.04) (2.84) (1.03) (0.52) (1.03) (1.08) Leverage 0.075*** 0.114 0.046 0.082*** 0.076*** 0.074*** (3.81) (1.56) (1.55) (2.86) (3.85) (3.73) Marketisation_index –0.005*** –0.001 –0.007*** –0.005** –0.005*** –0.005*** (–3.09) (–0.19) (–2.89) (–2.13) (–3.20) (–3.39) Independent_director –0.099* –0.159 –0.108 –0.095 –0.099* –0.100* (–1.70) (–1.12) (–1.01) (–1.21) (–1.70) (–1.72) Duality 0.001 0.044* –0.004 0.003 0.001 0.001 (0.12) (1.83) (–0.26) (0.19) (0.14) (0.13) Managerial_share 0.436*** 1.131*** 0.497*** 0.417*** 0.442*** 0.445*** (5.07) (3.61) (4.18) (3.59) (5.12) (5.22) China Journal of Accounting Studies 307 Constant 0.139*** 0.137*** 0.057 0.165*** 0.140*** 0.137*** 0.134*** (8.74) (4.93) (0.84) (3.31) (3.71) (4.90) (4.82) Year Yes Yes Yes Yes Yes Yes Yes Observations 10,494 10,494 593 3,520 6,381 10,494 10,494 Adjusted R 0.101 0.107 0.130 0.157 0.088 0.107 0.108 F value (model) 34.15 27.62 7.737 14.67 14.71 26.14 25.08 F value (ΔIncome* Income_decrease) (3) vs (5)=1.48 Note: For deﬁnitions of the variables, please refer to Table 2. ***, **, * indicate signiﬁcance levels of 1%, 5% and 10% respectively. 308 Liang et al. Table 6. Types of external auditors, agency costs and the expense stickiness. (1) High stake (2) Low stake (3) High (4) Low of major of major level of level of shareholders shareholders marketisation marketisation ΔIncome 0.332*** 0.241*** 0.271*** 0.317*** (7.67) (5.88) (7.37) (6.98) ΔIncome*Income_decrease –0.161 –0.137* –0.109 –0.217** (–1.27) (–1.83) (–1.38) (–2.38) ΔIncome*Income_decrease*Big4 0.011 0.690*** 0.079 0.717** (0.09) (2.98) (0.77) (2.36) Big4 0.011 0.064*** 0.030** 0.028 (0.77) (3.43) (2.49) (1.14) ΔIncome*Income_decrease 0.172 0.059 0.035 0.150 *Domestic_big10 (1.47) (0.71) (0.40) (1.20) Domestic_big10 0.018* 0.021* 0.010 0.032*** (1.77) (1.81) (1.04) (2.71) Two_year_decrease –0.048 –0.103*** –0.084*** –0.077*** (–1.54) (–4.64) (–3.97) (–2.84) GDP_growth –0.017*** –0.027*** –0.014** –0.027*** (–3.21) (–3.89) (–2.07) (–4.53) Employee_intensity –0.001 0.002 0.000 0.001 (–0.50) (0.65) (0.05) (0.34) Fixed_asset_intensity 0.004 0.002 0.003 0.002 (1.09) (0.72) (1.02) (0.54) Leverage 0.040 0.101*** 0.045* 0.103*** (1.42) (3.51) (1.82) (3.21) Marketisation_index –0.009*** –0.002 –0.010*** –0.005 (–4.09) (–0.93) (–2.75) (–1.04) Independent_director –0.102 –0.110 –0.068 –0.125 (–1.15) (–1.28) (–0.95) (–1.36) Duality 0.011 –0.004 0.008 –0.006 (0.68) (–0.30) (0.54) (–0.33) Managerial_share 0.490*** 0.510*** 0.516*** 0.151 (2.89) (5.16) (6.70) (0.68) Constant 0.174*** 0.114** 0.180*** 0.123** (4.38) (2.56) (3.59) (2.52) Year Yes Yes Yes Yes Observations 5,249 5,245 5,263 5,231 Adjusted R 0.150 0.078 0.126 0.095 F value (model) 15.24 12.79 18.28 11.89 F value (ΔIncome* (1) vs (2) = 6.46** (3) vs (4) = 4.01** Income_decrease) Note: For deﬁnitions of the variables, please refer to Table 2. ***, **, * indicate signiﬁcance levels of 1%, 5% and 10% respectively. Big Four International audit ﬁrms can reduce cost stickiness signiﬁcantly when the largest shareholder’s holding ratio is low. In addition, the F-value of the Chow test of the coefﬁcients of ΔIncome*Income_decrease*Big4 between the two groups is 6.46 and signiﬁcant at the 1% level. All of these results show that Big Four International audit ﬁrms play an active role when major shareholders hold fewer stakes and when agency costs are high. Similarly, the coefﬁcient of ΔIncome*Income_decrease*Big4 is not signiﬁcant in column (3) for ﬁrms in areas with higher levels of marketisation, indi- cating that Big Four International audit ﬁrms do not suppress cost stickiness signiﬁ- China Journal of Accounting Studies 309 cantly when ﬁrms are located in areas with well-developed markets. On the contrary, the coefﬁcient of ΔIncome*Income_decrease*Big4 is positive and signiﬁcant at the 5% level in column (4) for ﬁrms in areas with lower levels of marketisation, which means that Big Four International ﬁrms can signiﬁcantly restrict cost stickiness when ﬁrms are located in areas with less developed markets. In addition, the F-value of the Chow test of the coefﬁcients of ΔIncome*Income_decrease*Big4 is 4.01 and signiﬁcant at the 5% level. All these results suggest that Big Four International audit ﬁrms play a positive role in reducing cost stickiness when the market is not well developed and when agency cost is high. We do not ﬁnd evidence that the restrictive roles on cost stickiness are signiﬁcantly different between Big Ten Domestic and non–Big Ten Domestic audit ﬁrms. The above results together show in the role of Big Four International ﬁrms in mitigating agency costs and reducing cost stickiness changes under different circum- stances. 4.3. Robustness tests 4.3.1. Test using propensity score matching (PSM) So far, we have not ruled out the impact of self-selection on our results. For instance, companies with strict internal control may tend to choose audit services provided by Big Four International audit ﬁrms, whose services are of higher quality (e.g. Thornton & Moore, 1993). At the same time, these companies may exhibit lower cost stickiness due to strict internal control (Chen, Lu, & Sougiannis, 2012). If this is the case, our result that companies audited by Big Four International have lower cost stickiness is just a natural result of the companies’ choices. To address such concerns, we use the PSM method to further test our results. The PSM method can eliminate the effect of self-selection by matching the treated group (Big Four International-audited ﬁrms in this article) with the control group (non-Big Four International-audited ﬁrms) and is widely employed in related accounting and auditing research (e.g. Carcello, Vanstraelen, & Willenborg, 2009; Lawrence, Minutti-Meza, & Zhang, 2011). Speciﬁcally, the PSM test contains two stages. The ﬁrst stage is to grade based on choice tendency and the second stage is to match samples based on the grades gener- ated. The grading is based on a logit model whose dependent variable is an indicator variable of Big Four International audited ﬁrms (Big4). We choose independent vari- ables and control variables according to Wang et al. (2008). The grading model is Big4 ¼ a þ a SOE þ a Growth þ a Size þ a CROA þ a Leverage þ a CurrentR 0 1 2 3 4 5 6 þ a InventoryR þ a RecR þ RYear þ e (2) 7 8 where SOE is a dummy variable that equals one for state-owned companies and zero otherwise, Growth is the growth rate of operating revenue, Size is the natural logarithm of total assets, CROA is the proﬁt from operations divided by total assets, Lev is ﬁnan- cial leverage, CurrentR is the ratio of current assets to current liability, InventoryR is the ratio of inventory to total assets, and RecR is the ratio of accounts receivable to total assets. We also control for year ﬁxed effects. The ﬁrst-stage regression test shows that Size, Leverage, and other factors have signiﬁcant inﬂuence on whether a company hires a Big Four International audit ﬁrm. Matching by grades, we generate a total sam- ple of 1,084 observations, including both the treated and control groups, and we use this sample to conduct the second-stage regression. 310 Liang et al. Table 7 shows the regression results. Column (1) is based on full sample regression, while columns (2) to (5) report the results of subsamples grouped by largest share- holder’s holding ratio (columns (2) and (3)) or local marketisation levels (columns (3) and (4)). The results shows that the coefﬁcient of ΔIncome*Income_decrease*Big4 is positive and signiﬁcant at the 10% level in column (1), suggesting the restrictive role of Big Four International audit ﬁrms in cost stickiness. The coefﬁcient of ΔIncome*In- come_decrease*Big4 is not signiﬁcant in column (2) for ﬁrms with a high stake of major shareholders, but is positive and signiﬁcant at the 1% level in column (3) for ﬁrms with a low stake of major shareholders. Meanwhile, the F-value of the Chow test of the coefﬁcients of ΔIncome*Income_decrease*Big4 in the two columns is 3.86 and signiﬁcant at the 5% level. The coefﬁcient of ΔIncome*Income_decrease*Big4 is not signiﬁcant in column (4) for ﬁrms in areas with a well-developed market and is positive and signiﬁcant at the 10% level in column (5) for ﬁrms in areas with lower marketisa- tion levels. In addition, the F-value of the Chow test of the two coefﬁcients is 4.17 and signiﬁcant at the 5% level. These results are qualitatively the same as the results in pre- vious sections. The results in Table 7 show that using PSM does not change the out- come we found earlier. 4.3.2. Test on industrial listed companies Weiss (2010) uses only industrial listed companies to examine cost stickiness for the following two reasons: ﬁrst, the cost and proﬁt model of companies from a single industry allows for greater comparability. Second, industrial listed companies face more severe competition and using such a sample can better avoid the inﬂuence of monopo- listic prices. Therefore, we choose a subsample of industrial ﬁrms according to the industry classiﬁcation of the China Securities Market and Accounting Research (CSMAR) and conduct robust tests. Table 8 shows the regression results. Column (1) presents the full sample regres- sion, while columns (2) to (5) report the results of subsamples grouped by the largest shareholder’s holding ratio (columns (2) and (3)) or local marketisation levels (columns (3) and (4)). The results shows that the coefﬁcients of ΔIncome*Income_decrease*Big4 are positive and signiﬁcant at the 5% level in columns (1), (3), and (5), but are not sig- niﬁcant in column (2) for ﬁrms with a high stake of major shareholders or in column (4) for ﬁrms in areas with lower marketisation levels. These results support the positive role of the Big Four International audit ﬁrms in reducing cost stickiness and also the inference that the restrictive role varies with the largest shareholder’s holding ratio and local marketisation level. The results in Table 8 show that using a subsample of indus- trial ﬁrms does not change the outcome we found earlier. 4.3.3. Control for discretionary accruals Chen, Song, and Shi (2012) ﬁnd that companies may use a big bath to manage earnings to avoid future losses and such behaviour can lead to the overestimation of stickiness. Thus, we follow Dechow, Sloan, and Sweeney (1995) and use the adjusted Jones model to calculate discretionary accruals (DA). We then include DA and an interaction term ΔIncome*Income_decrease*DA in our model and re-estimate the model. The results are presented in Table 9. Column (1) presents the full sample regression, while columns (2) to (5) report the results of subsamples grouped by the largest shareholder’s holding ratio (columns (2) and (3)) or local marketisation levels (columns (3) and (4)). The China Journal of Accounting Studies 311 Table 7. Test using the PSM method. (1) Full (2) High stake of major (3) Low stake of major (4) High level of (5) Low level of sample shareholders shareholders marketisation marketisation ΔIncome 0.469*** 0.479*** 0.485*** 0.453*** 0.487*** (8.94) (5.27) (11.54) (7.62) (4.35) ΔIncome*Income_decrease –0.431*** –0.420** –0.629*** –0.163 –0.531** (–3.21) (–2.44) (–3.18) (–0.88) (–2.45) ΔIncome*Income_decrease*Big4 0.306* 0.144 0.753*** –0.035 0.848* (1.81) (0.92) (2.87) (–0.19) (1.96) Big4 –0.014 –0.038* 0.047 –0.010 –0.018 (–0.81) (–1.84) (1.63) (–0.51) (–0.56) Two_year_decrease 0.028 0.130* –0.181** 0.006 0.087 (0.50) (1.75) (–2.57) (0.12) (0.64) GDP_growth 0.001 –0.006 0.020 0.016 –0.015 (0.10) (–0.56) (1.42) (1.29) (–0.99) Employee_intensity –0.005 –0.004 –0.005 –0.001 –0.005 (–0.99) (–0.71) (–0.90) (–0.13) (–0.77) Fixed_asset_intensity 0.005 0.005 0.009 0.012** –0.001 (1.04) (0.92) (0.89) (2.05) (–0.12) Leverage 0.010 –0.003 –0.006 0.060 –0.017 (0.16) (–0.04) (–0.06) (0.89) (–0.13) Marketisation_index –0.006 –0.010* 0.004 –0.016 0.000 (–1.25) (–1.66) (0.51) (–1.63) (0.00) Independent_director –0.139 –0.015 –0.473** –0.160 –0.090 (–1.02) (–0.09) (–2.08) (–0.90) (–0.41) Duality 0.065* 0.051 0.103** 0.060 0.068 (1.77) (1.04) (2.03) (1.19) (1.60) Managerial_share –0.539 –3.056*** –0.391 –0.414 –0.817*** (–1.27) (–3.90) (–0.95) (–0.58) (–2.65) Constant 0.140** 0.158** 0.081 0.171 0.125 (2.22) (2.05) (0.73) (1.56) (0.98) Year Yes Yes Yes Yes Yes Observations 1,084 725 359 641 443 Adjusted R 0.179 0.143 0.298 0.254 0.101 F value (model) 6.908 4.905 13.89 8.414 45.14 F value (ΔIncome* (2) vs (3) = 3.86** (4) vs (5) = 4.17** Income_decrease) Note: For deﬁnitions of the variables, please refer to Table 2. ***, **, * indicate signiﬁcance levels of 1%, 5% and 10% respectively. 312 Liang et al. Table 8. Test on industrial listed companies. (1) Full (2) High stake of major (3) Low stake of major (4) High level of (5) Low level of sample shareholders shareholders marketisation marketisation ΔIncome 0.328*** 0.334*** 0.305*** 0.324*** 0.329*** (6.82) (5.54) (3.56) (5.13) (4.83) ΔIncome*Income_decrease –0.121 –0.007 –0.122 –0.140 –0.107 (–1.22) (–0.06) (–0.91) (–1.17) (–0.73) ΔIncome*Income_decrease*Big4 0.561** 0.301 0.700** 0.387 0.750** (2.47) (1.04) (2.33) (1.62) (2.48) Big4 0.018 –0.006 0.042** 0.026* 0.001 (1.26) (–0.35) (2.06) (1.73) (0.05) ΔIncome*Income_decrease 0.052 –0.011 0.063 0.054 0.041 *Domestic_big10 (0.47) (–0.09) (0.45) (0.42) (0.19) Domestic_big10 0.017* 0.006 0.024* 0.015 0.021 (1.93) (0.54) (1.75) (1.31) (1.51) Two_year_decrease –0.062*** –0.020 –0.083*** –0.070** –0.054 (–2.81) (–0.57) (–2.85) (–2.40) (–1.60) GDP_growth –0.023*** –0.021*** –0.024** –0.016* –0.028*** (–4.53) (–2.94) (–2.57) (–1.77) (–4.01) Employee_intensity –0.001 –0.004 0.001 –0.001 –0.002 (–0.34) (–1.05) (0.29) (–0.13) (–0.60) Fixed_asset_intensity 0.007 0.008 0.007 0.003 0.010 (1.41) (1.43) (1.07) (0.69) (1.28) Leverage 0.057** 0.019 0.089** 0.024 0.081** (2.32) (0.57) (2.40) (0.72) (2.19) Marketisation_index –0.004** –0.008*** –0.001 –0.013*** 0.000 (–1.97) (–2.80) (–0.18) (–2.83) (0.05) Independent_director –0.063 –0.089 –0.051 –0.100 –0.028 (–0.78) (–0.72) (–0.47) (–0.97) (–0.23) Duality –0.010 –0.001 –0.015 –0.003 –0.017 (–0.66) (–0.04) (–0.75) (–0.14) (–0.79) Managerial_share 0.329*** 0.305 0.382*** 0.441*** –0.060 (3.30) (1.43) (3.29) (4.65) (–0.38) China Journal of Accounting Studies 313 Constant 0.132*** 0.198*** 0.072 0.239*** 0.082 (3.46) (3.93) (1.18) (3.50) (1.40) Year Yes Yes Yes Yes Yes Observations 6,236 3,295 2,941 2,830 3,406 Adjusted R 0.106 0.128 0.086 0.138 0.089 F value (model) 14.33 8.998 7.523 11.65 7.891 F value (ΔIncome* (2) vs (3) = 0.88 (4) vs (5) = 1.00 Income_decrease) Note: For deﬁnitions of the variables, please refer to Table 2. ***, **, * indicate signiﬁcance levels of 1%, 5% and 10% respectively. 314 Liang et al. Table 9. Control for discretionary accruals. (1) Full (2) High stake of major (3) Low stake of major (4) High level of (5) Low level of sample shareholders shareholders marketisation marketisation ΔIncome 0.304*** 0.358*** 0.240*** 0.270*** 0.348*** (10.63) (8.56) (5.85) (7.36) (7.99) ΔIncome*Income_decrease –0.170*** –0.177 –0.130** –0.104 –0.237*** (–3.00) (–1.42) (–1.97) (–1.36) (–3.11) ΔIncome*Income_decrease*DA 0.773*** 0.488 0.846*** 0.350 1.322*** (3.84) (1.47) (3.37) (1.58) (3.84) DA –0.021 –0.091 0.016 0.019 –0.091 (–0.38) (–1.27) (0.18) (0.31) (–0.97) ΔIncome*Income_decrease*Big4 0.207 –0.020 0.645*** 0.070 0.651** (1.63) (–0.16) (2.96) (0.68) (2.44) Big4 0.029** 0.009 0.062*** 0.029** 0.022 (2.56) (0.62) (3.38) (2.44) (0.92) ΔIncome*Income_decrease 0.063 0.105 0.066 0.030 0.077 *Domestic_big10 (0.98) (0.85) (0.82) (0.36) (0.63) Domestic_big10 0.020*** 0.016 0.022* 0.010 0.029** (2.62) (1.58) (1.89) (1.06) (2.43) Two_year_decrease –0.085*** –0.050 –0.107*** –0.087*** –0.076*** (–5.01) (–1.62) (–4.80) (–4.06) (–2.86) GDP_growth –0.022*** –0.017*** –0.026*** –0.014** –0.027*** (–5.32) (–3.28) (–3.81) (–2.06) (–4.58) Employee_intensity –0.001 –0.002 0.000 –0.000 –0.001 (–0.31) (–0.90) (0.18) (–0.13) (–0.49) Fixed_asset_intensity 0.002 0.004 0.002 0.003 0.000 (1.12) (1.11) (0.80) (1.17) (0.09) Leverage 0.058*** 0.022 0.086*** 0.042* 0.065** (2.93) (0.77) (3.07) (1.70) (1.99) Marketisation_index –0.005*** –0.009*** –0.002 –0.010*** –0.005 (–3.58) (–4.16) (–1.05) (–2.78) (–1.06) Independent_director –0.098* –0.089 –0.119 –0.068 –0.120 (–1.70) (–1.00) (–1.43) (–0.95) (–1.34) China Journal of Accounting Studies 315 Duality 0.003 0.011 –0.002 0.008 –0.000 (0.27) (0.65) (–0.12) (0.55) (–0.02) Managerial_share 0.449*** 0.530*** 0.497*** 0.517*** 0.137 (5.27) (3.26) (4.95) (6.77) (0.60) Constant 0.143*** 0.175*** 0.129*** 0.183*** 0.141*** (5.29) (4.57) (2.99) (3.63) (3.07) Year Yes Yes Yes Yes Yes Observations 10,481 5,244 5,237 5,262 5,219 Adjusted R 0.117 0.162 0.086 0.128 0.114 F value (model) 25.48 15.74 13.02 17.16 13.65 F value (ΔIncome* (2) vs (3) = 6.71*** (4) vs (5) = 4.28** Income_decrease) Note: DA is discretionary accruals, which is calculated following Dechow et al. (1995). The other variables are deﬁned in Table 2. ***, **, * indicate signiﬁcance levels of 1%, 5% and 10% respectively. 316 Liang et al. results show that, after Da and the interaction term ΔIncome*Income_decrease*Da are controlled for, the coefﬁcient of ΔIncome*Income_decrease*Big4 is positive and signiﬁ- cant in column (1), while the coefﬁcient of ΔIncome*Income_decrease*Da is positive and signiﬁcant at the 1% level. The combined results indicate that Big Four International audit ﬁrms reduce cost stickiness mainly by inﬂuencing managers’ earnings management behaviour, since the introduction of discretionary accruals lowers the explanatory power of Big Four International audit ﬁrms for cost stickiness. Further subsample tests suggest that the coefﬁcients of ΔIncome*Income_decrease*Big4 are positive and signiﬁcant at the 1% and 5% levels, respectively, in columns (3) and (5), but are not signiﬁcant in col- umns (2) and (4). The results of Table 9 suggest that controlling discretionary accruals will not change our earlier outcomes. 4.3.4. Other robustness tests In addition, we carry out the following unreported tests. (1) Following Anderson et al. (2003), Chen, Lu, and Sougiannis (2012), and Sun and Liu (2004), we incorpo- rate the interactions of the four economic factors Two_year_decrease, GDP_growth, Employee_intensity, and Fixed_asset_intensity and stickiness (ΔIncome*Income_de- crease) or the interaction terms of all control variables and stickiness (ΔIncome*Inco- me_decrease) into the model. The results are qualitatively the same for both methods, with coefﬁcients of ΔIncome*Income_decrease*Big4 that are positive but not signiﬁ- cant in the full sample regressions and a signiﬁcant difference in the coefﬁcients of ΔIncome*Income_decrease*Big4 between different subsamples. (2) Following Chen, Song, and Shi (2012), we measure ﬁxed capital intensity by the ratio of ﬁxed assets to operating income. (3) We substitute Da with its absolute value. (4) We use yearly total income ranking to identify the Big Ten Domestic audit ﬁrms. (5) We choose the Big Five Domestic or the Big 20 Domestic instead of the Big Ten Domestic. None of the robustness tests changes the previous results qualitatively. 5. Conclusions and limitations Cost stickiness is an important problem for the issue of cost behaviour and an impor- tant measure of enterprise production and managerial efﬁciency. However, for a long time, related studies on cost stickiness have been limited and mainly focused on the impact of internal corporate governance mechanisms (Calleja et al., 2006; Chen, Lu, & Sougiannis, 2012). We introduce an important external governance mechanism, namely external audit, into this topic and propose that an external audit, which can mitigate information asymmetry and reduce internal agency costs through independent audit work, can play a restrictive role in a ﬁrm’s cost stickiness. Using China A-shares listed ﬁrms from 2002 to 2010, we test this hypothesis empirically. The results show that cost stickiness is less pronounced for ﬁrms audited by the Big Four International audit ﬁrms than for ﬁrms audited by non-Big Four Inter- national audit ﬁrms, but such a difference in cost stickiness is not found between ﬁrms audited by the Big Ten Domestic audit ﬁrms and by non-Big Ten Domestic audit ﬁrms. In addition, we ﬁnd that the restrictive role of the Big Four International is more pro- nounced when a ﬁrm’s largest shareholder’s holding ratio is lower or when a ﬁrm is located in an area with a lower marketisation level. The results are unchanged when we use PSM or industrial listed company subsamples or when we control for discre- tionary accruals. Our study enriches the literature on factors inﬂuencing cost stickiness China Journal of Accounting Studies 317 and indicates new incentives for the demand for high-quality audit services and implies the higher audit quality of Big Four International audit ﬁrms. Our paper also has the following limitations. First, there can be model misspeciﬁca- tion. (1) Our basic model follows Anderson et al. (2003), Chen, Lu, and Sougiannis (2012) and Sun and Liu (2004), but the missing quadratic term in the model can lead to econometric errors. (2) The selection of economic factors can cause problems. There are no conclusive standards for this issue and the domestic literature discusses this even less. This paper controls for the four most frequently selected factors, which are the indi- cator of two consecutive years of a decline in operating income, economic growth, human capital density, and ﬁxed capital intensity. However, this does not mean that the inﬂuence from the contract perspective and efﬁciency perspective is perfectly controlled for in our model. More variables reﬂecting these two perspectives may be gradually identiﬁed in the future. (3) There are other measures of cost stickiness. Weiss (2010) measures stickiness using quarterly operating ﬂuctuations, but seasonal sales will natu- rally affect the estimation of stickiness (e.g. the seasonal sales ﬂuctuation of electric appliances may not suggest a decline of sales). In addition, quarterly ﬁnancial statements are not mandatorily audited and may suffer more from earnings management. As a result, we do not use this method for robustness checks. Second, the self-selection prob- lem may still affect our results. Although we use PSM to match ﬁrms audited by Big Four International ﬁrms with control ﬁrms, we have to admit that such methods can only mitigate instead of rule out the impact of the self-selection problem and we should inter- pret the results with caution. These limitations can be addressed in future research. Acknowledgements This research is supported by grants from the National Natural Science Foundation of China (No. 71402198), the Beijing Municipal Commission of Education ‘Joint Construction Project’, the Beijing Municipal Commission of Education ‘Pilot Reform of Accounting Discipline Clustering’, ‘2011 Synergetic Innovation’ Key Project on ‘Development of Public Accounting Profession’ and ‘121 Talent Project for Young Doctoral Graduates Development 2014’ of the Central University of Finance and Economics. Notes 1. Chen, Lu, and Sougiannis (2012) provide direct evidence for this argument. They ﬁnd that cost stickiness has a positive correlation with corporate free cash ﬂow and this positive cor- relation can be weakened when the corporate governance level is higher. Calleja et al. (2006) ﬁnd that expense stickiness in France and Germany is higher than in the United States or Britain in their international study and they conjecture that the difference in corpo- rate governance is the main reason. 2. China is a large country where the levels of market development and government interven- tion vary signiﬁcantly across the 30 areas (provinces, autonomous regions or municipalities) (Jian & Wong, 2010). Fan, Wang, and Zhu (2010) offer an index that measures the develop- ment of marketisation of different areas in China. 3. In different years, it can be the Big Five International or the Big Eight International instead of Big Four International. For the sake of brevity, we use the term Big Four International for all conditions without differentiation. All these Big N international audit ﬁrms refer to the well-recognised top international audit ﬁrms. 4. The ﬁndings of these studies conﬂict for the following reasons, in addition to the difference in research focus (discretionary accruals, audit opinion types, ERC (earnings response coef- ﬁcient), and conservatism): ﬁrst, the sample period is short. Most sample periods are one year to three years and the longest are ﬁve years. The shorter the sample period, the more the variables of interest can ﬂuctuate, especially discretionary accruals. Second, the differ- 318 Liang et al. ence in the sample size of ﬁrms audited by Big Four International and non-Big Four Inter- national is not well addressed. Third, the self-selection problem is paid little attention. Our paper, instead, chooses a nine-year sample period (from 2002 to 2010) to conduct the empirical tests and methods such as propensity score matching (PSM) to mitigate the poten- tial problem. 5. The SME started in 2004 and GEM started in 2009. But the government stopped the IPO approval during the non-tradable share reform (2005–2006), so companies listed before 2007 are mainly on the main board. 6. Other reasons include the consideration of cost adjustment and managers’ optimism (Banker et al., 2011). Sun and Liu (2004) attribute these reasons to an opportunistic behaviour per- spective, a contract perspective, and an efﬁciency perspective. 7. However, their research design does not control for common variables, economic factor variables, or year or industry ﬁxed effects. 8. Calleja et al. (2006) mention external supervision theoretically, but do not conduct empirical research. 9. Such as overextending the number of employees (Williamson, 1963) and building luxurious ofﬁces. 10. Controlling earnings management is not the only way external governance works. If a com- pany’s performance worsens and managers still spend 1 million RMB to build luxurious ofﬁces, they can cover the adverse inﬂuence of such expenses on performance through earn- ings management. A capable and independent auditor can either correct the earnings man- agement during the auditing procedure, which reduces the degree of earnings management, or alert shareholders by signing non-standard audit opinions without correcting for earnings managements (in such cases, the degree of earnings management is unchanged). 11. Although many variables have been used in prior studies to measure agency costs, such as the administrative expense rate and free cash ﬂow, most of these variables might present the consequence of the whole corporate governance including auditing (Gul & Tsui, 1998; Li, 2007). 12. Our sample starts in 2002 because the concept of the Big Four International tends to have become stable with the voluntary surrender by Arthur Andersen of its licence to practice in the US. Also, the Chinese CPA Association started issuing its Comprehensive Evaluation of the Top 100 Accounting Firms in 2002, which provides a good reference for deﬁning the Big Ten Domestic audit ﬁrms. 13. The ﬁrm ranking is derived from yearly Comprehensive Evaluation of the Top 100 Account- ing Firms released by China CPA Association (http://www.cicpa.org.cn/Column/swszhpm). 14. These differences further support the necessity of using PSM as a robustness test. 15. Some later tests show that the sum of the coefﬁcients of ΔIncome*Income_decrease and ΔIncome*Income_decrease*Big4 is signiﬁcantly positive, meaning that there is an anti- stickiness phenomenon. This kind of result is mainly found in groups with lower largest shareholder’s shareholding ratio and groups with a lower market index. We conjecture that one possible reason is that the agency cost in this kind of company is higher. When sales decrease, managers trying to achieve performance goals may have to adopt real earnings management if their audit ﬁrms (Big Four International) effectively restrict them from con- ducting earnings management. In such a situation, excessive cost cutting can lead to an anti-stickiness phenomenon. Cohen and Zarowin (2010) and Zang (2012) both point out that the degree of real earnings management is higher when ﬁrms are audited by the Big Eight International. We do not ﬁnd an anti-stickiness phenomenon when PSM methods are used. 16. The medians of the largest shareholder’s holding ratio (Sh) and the market index (Marketi- sation_index) are 0.357 and 7.92, respectively. 17. We also estimate the model using 2×2 subsamples based on whether the ﬁrm is audited by Big Four International ﬁrms and whether the largest shareholder’s holding ratio (or market index) is higher than the median. The results show that the cost stickiness in groups when ﬁrms are audited by non-Big Four International ﬁrms and when the largest shareholder’s holding ratio (or when the market index is lower than the median) is the strongest. 18. The result is more signiﬁcant if we compare the Big Four International sample with the non-Big Ten Domestic sample (e.g. the control group does not include the Big Ten Domestic sample). China Journal of Accounting Studies 319 19. The idea is to mitigate the inﬂuence of earnings management in the form of a big bath, instead of the inﬂuence of the absolute value of earnings management. In accordance with this idea, we control for DA instead of its absolute value. In robustness tests, we control for the absolute value of discretionary accruals (AbsDA) and the interaction term ΔIncome*Income_decrease*AbsDA. 20. We think the restrictions on cost stickiness and earnings management are two parallel con- sequences. Big Four International ﬁrms can use different methods, such as restricting earn- ings management, signing non-standard audit opinions, and providing comparable ﬁnancial information of higher quality, to help shareholders make executive compensation and turn- over decisions to reduce managers’ opportunistic behaviour. Reducing earnings management is only one of these methods and, when the latter two methods are used, the degree of earn- ings management cannot completely explain the change in cost stickiness. 21. Unreported results are available from the authors upon request. 22. To address this problem, we also conduct subsample tests, but this method is convincing only in the test of the basic hypothesis. Further tests still rely on the basic model and may still be affected by potential model misspeciﬁcation (the comparison between subsamples with different ownership structure or different external environments is a Chow test of the cubic item). References Anderson, M., Banker, R., & Janakiraman, S. (2003). Are selling, general, and administrative costs ‘sticky’? Journal of Accounting Research, 41(1), 47–63. Ball, R., Robin, A., & Wu, J. S. (2003). Incentives versus standards: Properties of accounting income in four East Asian countries. Journal of Accounting and Economics, 36(1–3), 235–270. Banker, R. D., Byzalov, D., & Chen, L. T. (2013). Employment protection legislation, adjustment costs and cross-country differences in cost behavior. Journal of Accounting and Economics, 55(1), 111–127. Banker, R. D., Byzalov, D., & Plehn-Dujowich, J. M. (2011). Sticky cost behavior: Theory and evidence. Working Paper. Becker, C. L., Defond, M. L., Jiambalvo, J., & Subramanyam, K. R. (1998). The effect of audit quality on earnings management. Contemporary Accounting Research, 15(1), 1–24. Cai, C., Huang, Y. J., & Zhao, S. (2005). Empirical research on how audit quality inﬂuences earnings management: Evidence from manufacturing industry in Shanghai. Auditing Research, (2), 3–10 (in Chinese). Calleja, K., Steliaros, M., & Thomas, D. C. (2006). A note on cost stickiness: Some international comparisons. Management Accounting Research, 17(2), 127–140. Carcello, J. V., Vanstraelen, A., & Willenborg, M. (2009). Rules rather than discretion in audit standards: Going-concern opinions in Belgium. The Accounting Review, 84(5), 1395–1428. Chen, C. X., Lu, H., & Sougiannis, T. (2012). The agency problem, corporate governance, and the asymmetrical behavior of selling, general, and administrative costs. Contemporary Accounting Research, 29(1), 252–282. Chen, D. H., & Liang, S. K. (2010). Perks of managers, check-and-balance of stock ownership and its economic consequences: Empirical evidence from Chinese listed companies. Journal of Shanghai Lixin University of Commerce, (1), 19–27 (in Chinese). Chen, L., Song, L., & Shi, D. (2012). An empirical study on the upward estimation bias in cost stickiness: Evidence from Chinese listed companies. China Accounting Review, (1), 3–16 (in Chinese). China Securities Regulatory Commission and Ministry of Finance. (2001). The Regulation on the Temporary License for Foreign Accounting Firms to Conduct Business in Financial Listed Firms. (In Chinese.) Cohen, D. A., & Zarowin, P. (2010). Accrual-based and real earnings management activities around seasoned equity offerings. Journal of Accounting and Economics, 50(1), 2–19. Cohen, J., Krishnamoorthy, G., & Wright, A. M. (2002). Corporate governance and the audit pro- cess. Contemporary Accounting Research, 19(4), 573–594. DeAngelo, L. E. (1981). Auditor size and auditor quality. Journal of Accounting and Economics, 3(3), 183–199. 320 Liang et al. Dechow, P. M., Sloan, R. G., & Sweeney, A. P. (1995). Detecting earnings management. The Accounting Review, 70(2), 193–225. Defond, M. L., Wong, T. J., & Li, S. (2000). The impact of improved auditor independence on audit market concentration in China. Journal of Accounting and Economics, 28(3), 269–305. Dierynck, B., Landsman, W. R., & Renders, A. (2012). Do managerial incentives drive cost behavior? Evidence about the role of the zero earnings benchmark for labor cost behavior in private Belgian ﬁrms. The Accounting Review, 87, 1219–1246. Dopuch, N., & Simunic, D. A. (1980). The nature of competition in the auditing profession: A descriptive and normative view. In Buckley, J. and Weston, F. (Eds.), Regulation and the Accounting Profession. CA: Lifetime Learning Publications. Dou, W., Liu, X., & An, L. (2011). Ownership concentration, allocation of control rights and inefﬁcient investment in ﬁrms: Monitoring or colluding among multiple large shareholders? Journal of Management Sciences in China, 11,81–96 (in Chinese). Dye, R. (1993). Auditing standards, legal liability, and auditor wealth. Journal of Political Econ- omy, 101, 887–914. Fan, G., Wang, X. L., & Zhu, H. P. (2010). Marketization index for China’s provinces (2009), Economic Science Press (in Chinese). Fan, J., & Wong, T. J. (2005). Do external auditors perform a corporate governance role in emerging markets? Evidence from East Asia. Journal of Accounting Research, 43(1), 35–71. Fang, J. X., Hong, J. Q., & Li, R. S. (2004). Study on the inﬂuence factors of audit quality in listed companies in china: Discovery and enlightenment. Auditing Research, (6), 35–43. (in Chinese). Francis, J. R., Maydew, E. L., & Sparks, H. C. (1999). The role of big 6 auditors in the credible reporting of accruals. Auditing: A Journal of Practice & Theory, 18(2), 17–34. Gao, Q., & Wu, L. N. (2007). Large shareholder expropriation of funds and the auditor choice: Further evidence. Auditing Research, (5), 84–90 (in Chinese). Gong, Q. H., Liu, H. L., & Shen, H. H. (2010). The development of regional factor market, state holding and cost stickiness. China Accounting Review, (4), 431–445 (in Chinese). Gul, F. A., & Tsui, J. (1998). A test of the free cash ﬂow and debt monitoring hypotheses: Evi- dence from audit pricing. Journal of Accounting and Economics, 24, 219–237. Guo, Z. R. (2011). International big four accounting ﬁrms and high auditing quality: Evidence from China’s securities market. Auditing Research, (1), 98–107 (in Chinese). Hong, J. M., Xu, Y. D., & Li, Y. R. (2011). Information disclosure quality, controlling share- holder’s embezzlement and auditor selection. Auditing Research, (2), 107–112 (in Chinese). Jensen, M. C. (1986). Agency costs of free cash ﬂow, corporate ﬁnance, and takeovers. The American Economic Review, 76, 323–329. Jensen, M. C., & Meckling, W. H. (1976). Theory of the ﬁrm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3, 305–360. Jian, M., & Wong, T. J. (2010). Propping through related party transactions. Review of Account- ing Studies, 15(1), 70–105. Jiang, F. X., Huang, L., & Zhang, M. (2009). Product market competition, corporate governance and agency cost. The Journal of World Economy, (10), 46–59 (in Chinese). Jiang, R., Liu, X., & Liu, B. (2007). An empirical study on the governance effect of the exterior audit from Chinese listed companies: A view from CEO turnover. Journal of Finance and Economics, (11), 92–103 (in Chinese). Lawrence, A., Minutti-Meza, M., & Zhang, P. (2011). Can Big 4 versus non-Big 4 differences in audit-quality proxies be attributed to client characteristics? The Accounting Review, 86(1), 259–288. Lennox, C. S. (1999). Audit quality and auditor size: An evaluation of reputation and deep pock- ets hypotheses. Journal of Business Finance & Accounting, 26, 779–804. Li, S. X. (2007). Ownership, agency cost and agency efﬁciency. Economic Research Journal, (1), 102–113 (in Chinese). Li, W. A., Wang, X. H., & Wang, W. (2004). Earnings management and audit opinions: An empirical study based on non-operating earnings. Journal of Finance and Economics, (11), 126–135 (in Chinese). Liang, S. K. (2013). Ownership concentration and the viscosity of cost. Economic Research Jour- nal, (1), 144–155 (in Chinese). China Journal of Accounting Studies 321 Liu, F., & Xu, F. (2002). Risk based auditing, litigation risk and auditing quality. Accounting Research, (2), 21–27 (in Chinese). Liu, F., & Zhou, F. Y. (2007). Does size really matter? — A perspective of accounting conserva- tism test. Accounting Research, (3), 79–87 (in Chinese). Liu, J. H. (2009). Ownership, earning management and audit opinions. Auditing Research, (2), 32–39 (in Chinese). Liu, Q., & Yan, Z. Y. (2006). Accounting estimate and quality of independent auditing. Account- ing Research, (9), 82–88 (in Chinese). Ministry of Finance of the People’s Republic of China. (1993). Temporary Regulations on the Examination and Approval Procedures of Limited Liability Accounting Firms (in Chinese). Ministry of Finance of the People’s Republic of China. (1998). The Notice about Decoupling Accounting Firms which have Securities or Futures Related Business from their Linked Units (in Chinese). Moore, G., & Scott, W. (1989). Auditors’ legal liability, collusion with management. Contempo- rary Accounting Research, 5, 754–774. O’Keefe, T., & Westort, P. (1992). Conformance to GAAS reporting standards in municipal audits and the economics of auditing: The effects of audit ﬁrm size, CPA examination perfor- mance and competition. Research in Accounting Regulation, 6,39–77. Petersen, M. A. (2009). Estimating Standard errors in ﬁnance panel data sets: Comparing approaches. Review of Financial Studies, 22(1), 435–480. Qi, J. N., Chen, H. L., & Zhang, Y. (2004). Auditor size, brand, price and audit quality: Research on charging and quality of international big four Chinese audit marketing. Auditing Research, (3), 59–65 (in Chinese). Shleifer, A., & Vishny, R. (1986). Large shareholders and corporate control. Journal of Political Economy, 94, 461–488. Stulz, R. M. (1990). Managerial discretion and optimal ﬁnancing policies. Journal of Financial Economics, 26(1), 3–27. Sun, Z., & Liu, H. (2004). The expense ‘stickiness’ behavior of Chinese listed companies. Eco- nomic Research Journal, (12), 26–34 (in Chinese). Thornton, D. B., & Moore, G. (1993). Auditor choice and audit fee determinants. Journal of Business Finance & Accounting, 20, 333–349. Wan, S. Y., & Wang, H. J. (2011). Managerial self-interest, board governance and cost stickiness. Economy and Management, (3), 26–32 (in Chinese). Wang, J. C. (2011). Empirical study on the audit independence of top management turnover in the listed companies. Journal of Shanxi Finance and Economics University, (9), 116–124 (in Chinese). Wang, Q., Wong, T. J., & Xia, L. (2008). State ownership, the institutional environment, and auditor choice: Evidence from China. Journal of Accounting and Economics, 46(1), 112–134. Wang, X., & Zhang, W. G. (2005). Financial restatement and independent audit quality. Auditing Research, (3), 56–61 (in Chinese). Wang, Y. M., & Wang, P. (2006). Research on the difference in market reactions to auditor’s ﬁnancial statements between big-fours and non-big-fours. Auditing Research, (5), 49–56 (in Chinese). Watts, R. L., & Zimmerman, J. L. (1981). The markets for independence and independent audi- tors. Working Paper. Watts, R. L., & Zimmerman, J. L. (1983). Agency problems, auditing, and the theory of the ﬁrm: Some evidence. Journal of Law and Economics, 26, 613–633. Watts, R. L., & Zimmerman, J. L. (1986). Positive accounting theory. Englewood Cliffs, NJ: Prentice-Hall. Weiss, D. (2010). Cost behavior and analysts’ earnings forecasts. The Accounting Review, 85, 1441–1471. White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica, 48, 817–838. Williamson, O. E. (1963). Managerial discretion and business behavior. The American Economic Review, 53, 1032–1057. Wu, X. (2006). Accounting ﬁrm combination and quality control: The case of zhongtianqin. Accounting Research, (10), 79–85 (in Chinese). 322 Liang et al. Yang, D. M., Lin, B., & Wang, Y. C. (2009). Internal control, audit quality and large shareholder tunneling. Auditing Research, (5), 74–81 (in Chinese). Yu, P. (2007). Firm characteristics, the big-four and audit opinions. Auditing Research, (2), 53–60 (in Chinese). Yuan, H. Q., & Li, H. J. (2003). The organization form, scale and audit quality of accounting ﬁrms. Auditing Research, (1), 32–37 (in Chinese). Yue, H. (2006). Large shareholder tunneling and the monitoring role of auditor. China Accounting Review, (6), 59–68 (in Chinese). Zang, A. Y. (2012). Evidence on the trade-off between real activities manipulation and accrual-based earnings management. The Accounting Review, 87, 675–703. Zeng, Y., & Ye, K. T. (2005). Ownership structure, agency cost and external auditing demand. Accounting Research, (10), 63–70 (in Chinese). http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png China Journal of Accounting Studies Taylor & Francis http://www.deepdyve.com/lp/taylor-francis/external-auditor-types-and-the-cost-stickiness-of-listed-companies-kD0Fy0Tb7B

Loading next page...

References (70)

Jeffrey Cohen, G. Krishnamoorthy, A. Wright (2002)
Corporate Governance and the Audit Process
Contemporary Accounting Research, 19
(2010)
The development of regional factor market, state holding and cost stickiness
Dan Weiss (2010)
Cost Behavior and Analysts’ Earnings Forecasts
The Accounting Review, 85
Y. Heng (2006)
Large Shareholder Tunneling and the Monitoring Role of Auditor
China Accounting Review
(2007)
Firm characteristics, the big-four and audit opinions
Ming Jian, T. Wong (2010)
Propping through related party transactions
Review of Accounting Studies, 15
(2001)
The Regulation on the Temporary License for Foreign Accounting Firms to Conduct Business in Financial Listed Firms
(2006)
Accounting firm combination and quality control: The case of zhongtianqin
(2005)
Ownership structure, agency cost and external auditing demand
Bart Dierynck, W. Landsman, A. Renders (2012)
Do Managerial Incentives Drive Cost Behavior? Evidence about the Role of the Zero Earnings Benchmark for Labor Cost Behavior in Private Belgian Firms
The Accounting Review, 87
(2009)
Internal control, audit quality and large shareholder
(2006)
Accounting estimate and quality of independent auditing
R. Watts, J. Zimmerman (1981)
The Markets for Independence and Independent Auditors
C. Lennox (1999)
Audit Quality and Auditor Size: An Evaluation of Reputation and Deep Pockets Hypotheses
Journal of Business Finance & Accounting, 26
(2013)
Ownership concentration and the viscosity of cost
(2011)
International big four accounting firms and high auditing quality: Evidence from China’s securities market
Clara Chen, Hai-fan Lu, Theodore Sougiannis (2012)
The Agency Problem, Corporate Governance, and the Asymmetrical Behavior of Selling, General, and Administrative Costs
AAA 2009 Management Accounting Section (MAS) Meeting (Archive)
Wang Jin-chao (2011)
Empirical Study on the Audit Independence of Top Management Turnover in the Listed Companies
Journal of Shanxi Finance and Economics University
(1998)
The Notice about Decoupling Accounting Firms which have Securities or Futures Related Business from their Linked Units
Li Wei’an (2004)
Earnings Management and Audit Opinions:An Empirical Study Based on Non-operating Earnings
The Study of Finance and Economics
Qian Wang, T. Wong, Lijun Xia (2008)
State ownership, the institutional environment, and auditor choice: Evidence from China
Journal of Accounting and Economics, 46
F. Gang, W. Xiaolu, Zhang Liwen (2001)
Annual Report 2000 Marketization Index for China's Provinces
(2007)
Large shareholder expropriation of funds and the auditor choice: Further evidence
A. Lawrence, Miguel Minutti-Meza, Ping Zhang (2011)
Can Big 4 versus Non-Big 4 Differences in Audit-Quality Proxies Be Attributed to Client Characteristics?
The Accounting Review, 86
Daniel Thornton, Giora. Moore (1993)
AUDITOR CHOICE AND AUDIT FEE DETERMINANTS
Journal of Business Finance & Accounting, 20
(2009)
Ownership, earning management and audit opinions
(2002)
Risk based auditing, litigation risk and auditing quality
Mark Defond, T. Wong, Shuhua Li (1999)
The Impact of Improved Auditor Independence on Audit Market Concentration in China
USC Marshall School of Business Research Paper Series
Joseph Carcello, A. Vanstraelen, Michael Willenborg (2009)
Rules Rather than Discretion in Audit Standards: Going-Concern Opinions in Belgium
International Accounting eJournal
R. Dye (1993)
Auditing Standards, Legal Liability, and Auditor Wealth
Journal of Political Economy, 101
(2004)
Auditor size, brand, price and audit quality: Research on charging and quality of international big four Chinese audit marketing
Diogenis Baboukardos, Gunnar Rimmel (2016)
Positive Accounting Theory
(2007)
Does size really matter? — A perspective of accounting conservatism test
(2006)
Research on the difference in market reactions to auditor’s financial statements between big-fours and non-big-fours
(1992)
Conformance to GAAS reporting standards in municipal audits and the economics of auditing: The effects of audit firm size, CPA examination performance and competition
R. Banker, Dmitri Byzalov, J. Plehn-Dujowich (2010)
Sticky Cost Behavior: Theory and Evidence
Connie Becker, Mark Defond, J. Jiambalvo, K. Subramanyam (1998)
The Effect of Audit Quality on Earnings Management
Contemporary Accounting Research, 15
Michael Jensen (1986)
Agency Cost of Free Cash Flow, Corporate Finance, and Takeovers
Industrial Organization & Regulation eJournal
(2005)
Empirical research on how audit quality influences earnings management: Evidence from manufacturing industry in Shanghai
Amy Zang (2011)
Evidence on the Trade-Off between Real Activities Manipulation and Accrual-Based Earnings Management
Corporate Finance: Governance
(2003)
Incentives versus standards: Properties of accounting income in four East Asian countries
, 36
R. Watts, J. Zimmerman (1983)
Agency Problems, Auditing, and the Theory of the Firm: Some Evidence
The Journal of Law and Economics, 26
A. Ling (2011)
Ownership concentration,allocation of control rights and inefficient investment in firms: Monitoring or colluding among multiple large shareholders?
Journal of Management Sciences in China
Patricia Dechow, Richard Sloan, A. Sweeney (1994)
DETECTING EARNINGS MANAGEMENT
Accounting review: A quarterly journal of the American Accounting Association, 70
(2009)
Estimating Standard errors in finance panel data sets: Comparing approaches
(1993)
Temporary Regulations on the Examination and Approval Procedures of Limited Liability Accounting Firms
(2005)
Financial restatement and independent audit quality
R. Ball, Ashok Robin, Joanna Wu (2000)
Incentives Versus Standards: Properties of Accounting Income in Four East Asian Countries, and Implications for Acceptance of IAS
Chicago Booth Research Paper Series
René Stulz (1990)
Managerial discretion and optimal financing policies
Journal of Financial Economics, 26
(2011)
Information disclosure quality, controlling shareholder’s embezzlement and auditor selection
(2003)
The organization form, scale and audit quality of accounting firms
(2004)
Study on the influence factors of audit quality in listed companies in china: Discovery and enlightenment
Jere Francis, Edward Maydew, H. Sparks (1999)
The Role of Big 6 Auditors in the Credible Reporting of Accruals
Ear and Hearing, 18
Joseph Fan, T. Wong (2001)
Do External Auditors Perform a Corporate Governance Role in Emerging Markets? Evidence from East Asia
SEIN Corporate Governance & Accountability eJournal
(2009)
Product market competition, corporate governance and agency cost
R. Banker, Dmitri Byzalov, Lei Chen (2012)
Employment Protection Legislation, Adjustment Costs and Cross-Country Differences in Cost Behavior
Social Science Research Network
Giora. Moore, W. Scott (1989)
Auditors' legal liability, collusion with management, and investors' loss*
Contemporary Accounting Research, 5
D. Cohen, Paul Zarowin (2008)
Accrual-Based and Real Earnings Management Activities Around Seasoned Equity Offerings
New York University Stern School of Business Research Paper Series
Michael Jensen, W. Meckling (1976)
Theory of the Firm
(2011)
Managerial self-interest, board governance and cost stickiness
(2004)
The expense ‘stickiness’ behavior of Chinese listed companies
L. Deangelo (1981)
Auditor size and audit quality
Journal of Accounting and Economics, 3
Kenneth Calleja, Michael Steliaros, Dylan Thomas (2006)
A Note on Cost Stickiness: Some International Comparisons
Management Accounting Research, 17
(1980)
The nature of competition in the auditing profession: A descriptive and normative view
H. White (1980)
A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity
Econometrica, 48
L. Bin (2007)
An Empirical Study on the Governance Effect of the Exterior Audit from Chinese Listed Companies:A View from CEO Turnover
Journal of finance and economics
Li Shang-kun (2010)
Perks of Managers,Check-and-Balance of Stock Ownership and Its Economic Consequences:Empirical Evidence from Chinese Listed Companies
Journal of Shanghai Lixin University of Commerce
Andrei Shleifer, Robert Vishny (1986)
Large Shareholders and Corporate Control
Journal of Political Economy, 94
F. Gul, J. Tsui (1997)
A test of the free cash flow and debt monitoring hypotheses:: Evidence from audit pricing
Journal of Accounting and Economics, 24
Mark Anderson, R. Banker, S. Janakiraman (2003)
Are Selling, General, and Administrative Costs "Sticky"?
Journal of Accounting Research, 41

Publisher: Taylor & Francis
Copyright: © 2014 Accounting Society of China
ISSN: 2169-7221
eISSN: 2169-7213
DOI: 10.1080/21697213.2014.982004
Publisher site: See Article on Publisher Site

Abstract

China Journal of Accounting Studies, 2014 Vol. 2, No. 4, 294–322, http://dx.doi.org/10.1080/21697213.2014.982004 a b c Shangkun Liang *, Dong Chen and Xiaoli Hu China’s Management Accounting Research & Development Center/School of Accountancy, Central University of Finance and Economics, Beijing 100081, China; School of Economics and Management, Wuhan University, Wuhan, 430000, China; School of Business, National University of Singapore, Kent 119077, Singapore External auditing is an important method of corporate governance that can effec- tively monitor and restrict a manager’s opportunistic behaviour and reduce cost stickiness. We focus on Chinese A-shares listed companies from 2002 to 2010 to explore the governance role of external auditors in cost stickiness. We classify audit ﬁrms into three types: the Big Four International, the Big Ten Domestic, and the non-Big Ten Domestic. The results show that the cost stickiness of ﬁrms audited by the Big Four International is signiﬁcantly lower, while the cost stickiness difference between ﬁrms audited by the latter two types is not signiﬁcant. Further tests show that the restriction on cost stickiness for the Big Four International is more signiﬁ- cant when the largest shareholder’s shareholding ratio is lower or when the local market is less developed. These results indicate that the restriction placed by audit quality on cost stickiness varies with other corporate governance mechanisms. Our results are qualitatively unchanged under different robustness tests, including pro- pensity score matching. This study shows that Big Four International audit ﬁrms can better reduce cost stickiness, which provides evidence of the higher audit quality of these ﬁrms and extends the study of restrictive factors on cost stickiness to exter- nal corporate governance mechanisms. Keywords: Big Four International; cost stickiness; shareholding ratio; marketisation level; propensity score matching 1. Introduction External auditing is an indispensable part of internal corporate governance (Cohen, Krishnamoorthy, & Wright, 2002; Fan & Wong, 2005). The modern enterprise system features the separation of ownership and management. This mechanism might improve the efﬁciency of capital and human talents. However, it also inevitably leads to conﬂict between principals and agents (Jensen & Meckling, 1976). It is possible that agents (managers) with information superiority would harm the interest of principals (share- holders) to maximise their own interests. External audits can mitigate information asymmetry between contracting parties and reduce internal agency cost through inde- pendent audit work (Watts & Zimmerman, 1983). Many Chinese studies focus on the choice and impact of external auditors, including constraints imposed by external audi- tors on earnings management and ﬁnancial restatement (Cai, Huang, & Zhao, 2005; Li, Wang, & Wang, 2004; Liu, 2009; Liu & Yan, 2006; Wang & Zhang, 2005), and on related-party transactions (Gao & Wu, 2007; Hong, Xu, & Li, 2011; Yang, Lin, & *Corresponding author. Email: jevonacc@gmail.com Paper accepted by Xi Wu. © 2014 Accounting Society of China China Journal of Accounting Studies 295 Wang, 2009; Yue, 2006; Zeng & Ye, 2005). There is also a constraint through the governance role of the external auditors in the turnover of chief executive ofﬁcers (CEOs) (Jiang, Liu, & Liu, 2007; Wang, 2011). However, there are fewer studies on the constraints imposed by external auditing on managerial behaviour (Jiang et al., 2007) in this strand of literature. Such studies as have been carried out focus mainly on the quality of information disclosure, which is the ﬁnal output of the information dis- closure process. This study argues that external audit governance is effective not only in the ﬁnal part of information disclosure, but also during operational and production procedures. To explore this possibility, we examine the inﬂuence of external audit on cost stickiness. Cost stickiness refers to the phenomenon where costs decrease less when business activity falls than they increase when business activity rises (Anderson, Banker, & Janakiraman, 2003). Anderson et al. (2003) tested this phenomenon for the ﬁrst time in the ﬁeld of ﬁnance and accounting, using US data. We look at the association between external audit and cost stickiness in China for the following three reasons. First, this study allows us to explore the mechanism of for- mation of cost stickiness from an external perspective. The literature on cost stickiness is mainly based on an internal perspective. Anderson et al. (2003), Calleja, Steliaros, and Thomas (2006), Chen, Lu, and Sougiannis (2012), and Sun and Liu (2004) point out that managerial opportunistic behaviour is one of the primary causes of cost sticki- ness and they mainly focus on internal factors such as the structure of the board of directors. However, our study extends the investigation of the causes of cost stickiness to external governance mechanisms and looks at the role of external auditors for the ﬁrst time, which contributes to the current literature. Second, it explores and shows the incentive for the demand for high-quality external audits, given China’s institutional background. Ball, Robin, and Wu (2003) argue that the large number of related-party transactions in East Asian countries leads to a lack of motivation for listed ﬁrms to pro- vide high-quality accounting information. Defond, Wong, and Li (2000) also point out that, under government regulation, Chinese listed companies lack demand for high- quality audit services. However, the environment of Chinese listed ﬁrms has changed with the launch of non-tradable share reform, the development of marketisation, and the move to public listing of many non-state-owned ﬁrms. Since the conﬂict between shareholders and managers is becoming increasingly severe, the need to monitor man- agers can become one source of demand for high-quality audits. Third, it provides evi- dence for the debate on the audit quality of the Big Four International ﬁrms. Most international studies agree that the audit services of the Big Four International ﬁrms are of higher quality (Becker, Defond, Jiambalvo, & Subramanyam, 1998; Francis, Maydew, & Sparks, 1999; Lennox, 1999). However, although the same opinion is dominant in Chinese studies, controversy still exists (Guo, 2011; Liu & Xu, 2002; Liu & Zhou, 2007; Qi, Chen, & Zhang, 2004; Wang & Wang, 2006;Yu, 2007). We focus on the governance role of the Big Four International ﬁrms and the Big Ten Domestic audit ﬁrms in China in order to provide evidence for this debate from a new perspective. When there is a business boom, optimistic managers are likely to increase expenses (especially unproductive expenditures) drastically to enhance personal utility or to con- trol more resources (Chen, Lu, & Sougiannis, 2012). When there is a business reces- sion, however, managers may not be willing to decrease such expenses or may just decrease them slightly. Such behaviour results in high cost stickiness; that is, expenses decrease more slowly when revenue falls than expenses increase when revenue rises by the same amount. To the extent that high-quality external audits can effectively regulate 296 Liang et al. the corporate ﬁnancial process and constrain managers’ opportunistic behaviour (Watts & Zimmerman, 1983), we suggest high-quality external audits may help to reduce cost stickiness. Using Chinese A-shares listed companies from 2002 to 2010, we test the relationship between high quality external audit and cost stickiness. We classify audit ﬁrms into three types: the Big Four International, the Big Ten Domestic, and the non- Big Ten Domestic. The results show that the cost stickiness is signiﬁcantly lower in listed companies audited by the Big Four International, while there is no signiﬁcant cost stickiness difference between ﬁrms audited by the latter two types. In addition, the restriction on cost stickiness associated with the Big Four International is much more signiﬁcant when the largest shareholder’s shareholding ratio is lower or when the local market is less developed, which indicates that the audit quality’s restriction on cost stickiness varies with other corporate governance mechanisms. This study reveals that the Big Four International are associated with cost stickiness and extends the study on cost stickiness factors to external governance mechanisms. This paper contributes to the literature in the following three ways. First, it enriches the literature that explores factors associated with cost stickiness and sheds light on how external audit quality interacts with factors identiﬁed in prior literature. This paper investigates empirically for the ﬁrst time how auditor supervision, an external gover- nance method, impacts cost stickiness and ﬁnds that the Big Four International consid- erably inhibit cost stickiness. In addition, we ﬁnd that the impact of audit quality on cost stickiness varies with changes in other corporate governance mechanisms, such as ownership structure and external environment, which helps to understand cost stickiness factors and how these other factors affect cost stickiness. Second, it shows motivation for the demand for high-quality auditing in China. We ﬁnd the Big Four International ﬁrms are associated signiﬁcantly with cost stickiness (e.g. when the shareholding ratio of major shareholders is lower). Such a restrictive effect on cost stickiness could effec- tively increase the beneﬁts of hiring audit ﬁrms with higher audit quality and add to the demand for high-quality audit services. Third, we provide evidence of the quality differences between different audit ﬁrms from a new perspective. We show that while the Big Four International are signiﬁcantly associated with cost stickiness, the Big Ten Domestic ﬁrms do not have a similar impact. Our results are consistent with the argu- ment that the Big Four International produce higher-quality audit work. The rest of the paper is structured as follows. Section 2 presents the institutional background, literature review and hypothesis development. Section 3 describes the research design. Section 4 shows the empirical test results. Section 5 concludes and dis- cusses potential limitations. 2. Institutional background, literature review, and hypothesis development 2.1. Institutional background The development of external auditing in China is closely related to politics and the economy. After China’s reform and opening up (after 1978), the CPA (Certiﬁed Public Accountants) system was restored and the audit business redeveloped (Liu & Xu, 2002). According to the Temporary Regulations on the Examination and Approval Pro- cedures of Limited Liability Accounting Firms passed by the Ministry of Finance in 1993, early accounting ﬁrms were mostly established relying on government agencies, such as the local Finance Departments of the different provinces. This situation remained the same until April 1998, when the Ministry of Finance released The Notice about Decoupling Accounting Firms Which Have Securities or Futures Related China Journal of Accounting Studies 297 Business from Their Linked Units, which launched the decoupling reform. By the end of 1999, the decoupling reform was mostly accomplished. However, domestic account- ing ﬁrms that originally relied on government agencies still had many inherent deﬁcien- cies. Along with several serious incidents of fraud in China’s capital market, market attention to audit quality gradually grew, initiating the demand for the audit work of the original Big Five International ﬁrms. In 2001, the China Securities Regulatory Commission together with the Ministry of Finance released The Regulation on the Tem- porary License for Foreign Accounting Firms to Conduct Business in Financial Listed Firms. In the same year, the two regulatory departments authorised temporary licences to the Big Four (originally the Big Five) international accounting ﬁrms and required that listed ﬁrms in the banking, security, and insurance industry be audited by both domestic and international ﬁrms. Similar administrative regulations requiring certain types of ﬁrms to be audited by international audit ﬁrms were also implemented after- wards (Liu & Xu, 2002; Liu & Zhou, 2007). In 2001 China signed an agreement to open up the CPA service market when it joined the World Trade Organization (Qi et al., 2004), which led to the complete opening of the auditing market to the Big Four International. Over the past ten years, the Big Four International ﬁrms have developed rapidly in the audit market in China. Their total income accounted for more than 55% of the total income of China’s top 100 accounting ﬁrms at the end of 2007. After 2007, as a large number of small companies went public in the SME (small and medium sized enterprises) and GEM (growth enterprises market) boards, the income share of the Big Four International declined but still surpassed 30% of the total income of the top 100 accounting ﬁrms (according to data from The Comprehensive Evaluation of Top 100 Accounting Firms in China published by the Chinese Association of Certiﬁed Public Accountants from 2003 to 2013). In addition, the Big Four International audit ﬁrms are widely recognised as having professional advantages in human capital, accu- mulative knowledge, and non-audit services (Qi et al., 2004; Wang & Wang, 2006). 2.2. Literature review Anderson et al. (2003) were the ﬁrst in the ﬁeld of ﬁnance and accounting to propose the concept of cost stickiness, which refers to the phenomenon where costs decrease less when business activity falls than they increase when business activity rises. The authors use a sample of US-listed companies to test this theory empirically. They ﬁnd that selling, general, and administrative (SG&A) costs increase by 0.55% when sales revenue increases by 1% but only decrease by 0.35% when sales revenue decreases by 1%, which provides preliminary evidence of cost stickiness. Numerous studies further conﬁrm the existence of cost stickiness (Calleja et al., 2006; Chen, Lu, & Sougiannia, 2012; Dierynck, Landsman, & Renders, 2012) and they focus more on the causes of cost stickiness. Among these causes, managers’ opportu- nistic behaviour is considered to be crucial (Banker, Byzalov, & Plehn-Dujowich, 2011; Sun & Liu, 2004) and empirical evidence is relatively abundant. Calleja et al. (2006) compare the degree of cost stickiness in four countries. They point out that weaker cor- porate governance accounts for the higher cost stickiness in France and Germany than in the United States and the United Kingdom. Chen, Lu, and Sougiannis (2012) further examine the effect of agency costs between stockholders and managers on cost sticki- ness. They ﬁnd that cost stickiness is highly positively related to the incentive for man- agerial empire building (such as free cash ﬂow, executive tenure, expectation to leave, and compensation structure). They also ﬁnd that strong corporate governance can 298 Liang et al. mitigate this positive association. Dierynck et al. (2012) provide evidence that executives’ incentive to conduct earnings management affects labour cost stickiness. They ﬁnd that, to meet the goal of proﬁt, management tends to manipulate labour costs. Apart from managers’ opportunistic behaviour, adjustment cost and management opti- mistic expectations can also cause cost stickiness. Regarding adjustment cost, Anderson et al. (2003) propose that ﬁrms with large proportions of physical assets and human capital tend to have greater cost stickiness, since the adjustment costs of these two assets are higher. Banker, Byzalor, and Chen (2013) ﬁnd that the power of a country’s labour union has a signiﬁcant inﬂuence on the adjustment costs of human resource. If a county’s labour union is more powerful, the adjustment cost of human resources will be greater and labour cost stickiness will be correspondingly higher. Regarding man- agement optimistic expectations, Anderson et al. (2003) suggest that when the macro- economy grows rapidly, management will be optimistic about the future, even after a short-term decline in revenue, so cost stickiness will be greater. Similarly, Banker et al. (2011) ﬁnd that companies in growing industries have greater cost stickiness. In terms of Chinese domestic studies, Sun and Liu (2004) ﬁnd that cost stickiness also exists in China’s listed companies, but macroeconomic growth and capital inten- sive cannot explain it well. Therefore, managers’ opportunistic behaviour may be a bet- ter explanation. Wan and Wang (2011) ﬁnd that CEO non-duality and a higher proportion of independent directors can effectively reduce cost stickiness. Chen, Song, and Shi (2012) propose that controlling for managers’ incentives to manage earnings, as in a ‘big bath’, can drastically decrease the estimated cost stickiness. Liang (2013) ﬁnds that cost stickiness is negatively associated with the level of ownership concentration. Given the ﬁndings of the foreign and domestic literature, there have been many studies on the impact of managers’ opportunistic behaviour on cost stickiness. How- ever, these studies mainly focus on the effect of internal corporate governance mecha- nisms rather than external governance mechanisms. Based on the studies above, we further include audit supervision, an important external governance factor, in this line of study and examine the effect of external audit quality on the cost stickiness of listed companies. 2.3. Hypothesis development In this paper, we seek to develop the argument that that high-quality external audits can reduce listed companies’ cost stickiness. Managers’ opportunistic behaviour is a main cause of cost stickiness (Banker et al., 2011). In modern corporations, with the separation of management and ownership, minority shareholders have to pay large amounts to obtain comprehensive information about companies’ operation and develop- ment, while managers do not (Jensen & Meckling, 1976). In addition, since executive incentive contracts cannot completely reconcile shareholders’ and managers’ interests, managers pervasively have an ‘empire building’ incentive. Managers tend to over- expend ﬁrm size and spend a great amount of resources on personal gains instead of improving company value (Jensen, 1986; Stulz, 1990). When business performance declines, managers’ optimal behaviour should be to cut down costs and keep them in accordance with activities. However, managers with strong opportunistic incentives will not adjust or only slowly adjust costs when sales decline (Chen, Lu, & Sougiannis, 2012). For example, to pursue personal prestige and rights, managers hire large num- bers of employees, frequently update ofﬁce equipment, and incur many other expenses. China Journal of Accounting Studies 299 In such a situation, costs decrease less when business activity falls than costs increase when business activity rises and we regard this asymmetric cost change as cost stickiness. We propose that high-quality external audits can decrease managers’ opportunistic behaviour and further help decrease cost stickiness. Becker et al. (1998) and Francis et al. (1999) point out that high-quality external audits have a great inﬂuence on corpo- rate behaviour, such as standardising the ﬁnancial procedures of listed companies, improving information transparency, and reducing ﬁnancial fraud and earnings manage- ment. High-quality external audits can identify and correct management fraud and earnings management (Qi et al., 2004). They can also alert shareholders by providing non-standard audit opinions (Jiang et al., 2007), which indirectly limits managers’ opportunistic behaviour. Under weak audit supervision, managers can package costs or conduct earnings management to hide the negative impact of such costs on company performance. However, under strong audit supervision, it is difﬁcult to hide the nega- tive impact. Considering the difﬁculties of conducting ﬁnancial fraud or earnings man- agement, managers will decrease their opportunistic behaviour under strong audit supervision and cost stickiness will decrease. In addition, through standardising ﬁnan- cial procedures, high-level external audits can also improve the quality of ﬁnancial report information to help shareholders compare the information with other companies in the same industry accurately and in a timely manner and identify the relation of abnormal revenues and abnormal costs. Shareholders can thus make better decisions about executive compensation and replacement and strengthen control on management. Previous studies show that Big Four International ﬁrms, with high-level specialisa- tion and independence, can represent high-quality external auditing. Dopuch and Simu- nic (1980) ﬁnd that big accounting ﬁrms tend to undergo more specialised training, stricter certiﬁcation, and peer review. O’Keefe and Westort (1992) ﬁnd that big accounting ﬁrms invest more in knowledge. When the size of auditor grows, the quasi- rent for each client will decrease. Then the probability of an audit ﬁrm behaving oppor- tunistically will decrease, so the audit ﬁrm is more independent (DeAngelo, 1981). Dye (1993), Moore and Scott (1989) and Watts and Zimmerman (1981, 1986) demonstrate that big accounting ﬁrms are more independent from the perspective of the brand repu- tation mechanism and deep pocket theory. Greater ability helps audit ﬁrms identify and correct ﬁnancial procedure weakness and restrict managers’ earnings management. If the accounting ﬁrms are more independent, they are more likely to provide objective and fair audit reports and can increase deterrence to managers. Compared with domes- tic audit ﬁrms, Big Four International ﬁrms have obvious advantages due to their his- tory, size, specialisation, market share, and so forth, which ensure their higher audit quality (Qi et al., 2004). In addition, Wang, Wong, and Xia (2008) ﬁnd that, in China, government intervention can reduce audit quality, especially the quality of small local audit ﬁrms. We can infer from the previous analysis that, compared with domestic ﬁrms, the Big Four International provide better audit services and companies audited by them tend to have lower degrees of cost stickiness. Similarly, according to the quasi-rent and deep pocket theories, the Big Ten Domestic audit ﬁrms should have higher audit quality than other domestic ﬁrms and their clients’ cost stickiness should therefore be lower than that of the clients of other domestic ﬁrms. The above discus- sion leads to our ﬁrst hypothesis. H1a: Compared with listed companies audited by domestic audit ﬁrms, companies audited by the Big Four International have a lower level of cost stickiness. 300 Liang et al. H1b: Compared with listed companies audited by non-Big Ten Domestic audit ﬁrms, companies audited by the Big Ten Domestic have a lower level of cost stickiness. Our prior analysis suggests high-quality external audit can mitigate cost stickiness by reducing agency costs between stockholders and managers. However, external audit is only part of corporate governance and other governance mechanisms can also reduce agency costs and control managers’ opportunistic behaviour. If other mechanisms can successfully reduce agency costs, external audit may not be the dominant factor in the supervision, but if other mechanisms fail, external audit can be more important. Speciﬁcally, we analyse the interaction between different forms of external gover- nance regarding two aspects: ownership structure and the external environment. The largest shareholder’s holding ratio and marketisation level of the province where a ﬁrm’s headquarter is located represent two important dimensions of external governance mechanisms and are relatively exogenous to the Big Four International indicator. Since these two factors can vary cross-sectionally, the degree of corporate agency prob- lem can vary accordingly. If the largest shareholder possesses a higher proportion of shares, it will monitor the managers more actively (Shleifer & Vishny, 1986). The pro- portion of shares held by the largest shareholder is negatively associated with agency costs. Chen and Liang (2010) ﬁnd that non-pecuniary compensation is attenuated when the largest shareholder holds a higher proportion of shares. Dou, Liu, and An (2011) ﬁnd that large shareholders’ complete control of a ﬁrm can prevent inefﬁcient invest- ments. In addition, when the market is highly developed, listed companies will face more competition and less government intervention and investors can identify manag- ers’ efforts accurately and in a timely fashion. As a result, a high level of marketisation leads to low agency costs. Li (2007) ﬁnds that ﬁrms’ SG&A rate and efﬁciency loss both decrease when market competition increases. Jiang, Huang, and Zhang (2009) ﬁnd that administrative expense rates decrease and asset turnover increases in highly com- petitive product markets. When the largest shareholder holds a lower proportion of shares and the market level is lower, agency costs are relatively higher and an external audit can play a more important role in reducing cost stickiness. The above discussion leads to our second hypothesis. H2a: When the largest shareholder holds a lower proportion of shares, the negative associ- ation between being audited by the Big Four International or Big Ten Domestic and cost stickiness is more pronounced. H2b: When the market level is lower, the negative association between being audited by the Big Four International or Big Ten Domestic and cost stickiness is more pronounced. 3. Research design 3.1. Sample selection Our initial sample starts with all A-shares listed ﬁrms in the Shanghai and Shenzhen Stock Exchanges from 2002 to 2010. We then delete 1670 observations that were listed for less than two years and 133 observations in the ﬁnancial industry. We also delete observations whose change in operating revenue or expense from the previous year is lower than 0.5% or higher than 99.5%, which includes 723 observations (Chen, Lu, & Sougiannis, 2012). Finally, we delete 534 observations with missing controlling variables. Our ﬁnal sample includes 10,494 observations. Table 1 shows the sample selection procedures. China Journal of Accounting Studies 301 Table 1. Sample selection process. Listed less Firms with extreme Initial than 2 Financial changes in income or Missing Remaining Year observations years companies expense (0.5%) data observations 2002 1,211 150 7 64 56 934 2003 1,278 138 7 71 43 1,019 2004 1,377 167 8 88 34 1,080 2005 1,366 115 10 88 55 1,098 2006 1,421 81 10 98 65 1,167 2007 1,537 192 14 107 52 1,172 2008 1,611 203 17 47 72 1,272 2009 1,703 176 30 80 82 1,335 2010 2,050 448 30 80 75 1,417 Total 13,554 1,670 133 723 534 10,494 3.2. Research design and variable deﬁnitions Following Anderson et al. (2003), Chen, Lu, and Sougiannis (2012) and Sun and Liu (2004), we establish the following model to test our hypothesis: DCost ¼ a þ a DIncome þ a DIncome Income decrease þ a DIncome 0 1 2 3 Income decrease Big4 þ a Big4 þ a DIncome Income decrease 4 5 Domestic big10 þ a Domestic big10 þ RControl þ e (1) where ΔCost and ΔIncome represent changes in expense and in operating income, respectively. We ﬁrst calculate the ratio of expenses in year t to expense in year t – 1 and the ratio of the operating income in year t to that in year t – 1 and then take the natural logarithm of these two ratios. Speciﬁcally, expense is the sum of operating cost and general cost; Income_decrease is a dummy variable for income decline, which equals one if the operating income in year t was lower than that in year t – 1 and zero otherwise; Big4 is an indicator variable for Big Four International audit ﬁrms that equals one if the listed company was audited by Ernst & Young, KPMG, Deloitte, or PwC in year t and zero otherwise; and Domestic_big10 is a dummy variable of Big Ten Domestic audit ﬁrms that equals 1 if the ﬁrm was audited by a Big Ten Domestic audit ﬁrm and zero otherwise. Following Anderson et al. (2003), Chen, Lu, and Sougiannis (2012) and Sun and Liu (2004), our control variables include an indicator variable for Income decline for two consecutive years (Two_year_decrease), economic growth (GDP_growth), human capital intensity (Employee_intensity), ﬁxed capital intensity (Fixed_asset_intensity), ﬁnancial leverage (Leverage), the market index (Marketisation_index), proportion of independent directors (Independent_director), CEO duality (Duality), and proportion of managerial shareholdings (Managerial_share). Table 2 shows the deﬁnitions and explanations of the main variables. To reduce the effect of extreme values, all continuous variables are winsorized at the 0.5% level. We control for year ﬁxed effects and clustered observations by ﬁrms (Petersen, 2009)to mitigate the effect of serial correlation. We also report robust t-values to adjust for het- eroscedasticity (White, 1980). These methods improve the robustness of our empirical results. If cost stickiness exists, the regression coefﬁcient a in model (1) should be signiﬁ- cantly negative. If H1a holds, we would observe that a in model (1) is signiﬁcantly 3 302 Liang et al. Table 2. Variable deﬁnitions. Type Variables name Variables symbol Variables deﬁnition Dependent Changes in the ΔCost Natural logarithm of the ratio of the variable costs expense in year t to the expense in year t–1 Independent Changes in ΔIncome Natural logarithm of the ratio of the variables operating income operating income in year t to the operating income in year t–1 Income declining Income_decrease Dummy variable that equals 1 if operating income in year t is lower than that in year t–1, and equals 0 otherwise The Big Four Big4 Dummy variable that equals 1 if the International ﬁrm is audited by Big Four International, 0 otherwise The Big Ten Domestic_big10 Dummy variable that equals 1 if the Domestic ﬁrm is audited by Big Ten Domestic, 0 otherwise Control Income declining Two_year_decrease Dummy variable that equals 1 if the variables for two operating income declines for two consecutive years consecutive years, and equals 0 otherwise Economic growth GDP_growth GDP growth for year t Human capital Employee_intensity Number of employees/operating revenue intensity (in millions) Fixed capital Fixed_asset_intensity Total assets/operating revenue intensity Financial leverage Leverage Debts/assets Marketisation Marketisation_index Marketisation index (Fan et al., 2010) index Proportion of Independent_director Proportion of independent directors independent directors CEO duality Duality Dummy variable that equals 1 if the CEO and Chairman are the same one, otherwise 0 Proportion of Managerial_share Proportion of shares held by managers management shareholdings positive. If H1b holds, a should be signiﬁcantly positive. If H2 holds, we expect that a and a in model (1) exhibit signiﬁcant differences between groups with different 3 5 ownership structures or local marketisation levels. 3.3. Descriptive statistics and correlation coefﬁcients Table 3 reports the descriptive statistics of the main variables. Panel A shows the descriptive statistics for the full sample. The means of ΔCost and ΔIncome are 0.133 and 0.145, respectively, and the ratio of observations with an income decrease (mean of Income_decrease) is 25.6%. These statistics are close to those reported by Gong, Liu, and Shen (2010). The proportion of ﬁrm-years audited by Big Four International (mean of Big4) is 5.65%, which is close to the result of Guo (2011), Wang and China Journal of Accounting Studies 303 Table 3. Descriptive statistics. Panel A: Full sample Variables symbol Mean Lower quartile Median Upper quartile Standard deviation ΔCost 0.133 –1.499 0.132 1.752 0.375 ΔIncome 0.145 –1.525 0.141 2.151 0.391 Income_decrease 0.256 0.000 0.000 1.000 0.436 Big4 0.057 0.000 0.000 1.000 0.231 Domestic_big10 0.335 0.000 0.000 1.000 0.472 Two_year_decrease 0.096 0.000 0.000 1.000 0.295 GDP_growth 10.726 9.100 10.100 14.200 1.618 Employee_intensity 2.717 0.023 1.700 28.418 3.503 Fixed_asset_intensity 2.717 0.304 1.794 36.022 3.630 Leverage 0.506 0.062 0.515 0.957 0.186 Marketisation_index 8.060 0.630 7.920 11.710 2.202 Independent_director 0.346 0.000 0.333 0.600 0.073 Duality 0.096 0.000 0.000 1.000 0.295 Managerial_share 0.005 0.000 0.000 0.273 0.028 Panel B: Sub-sample Big Four Big Four Big Ten Non-Big Ten International–Non- International Domestic Domestic Big Ten Domestic Variables symbol Mean Median Mean Median Mean Median Mean Median ΔCost 0.145 0.127 0.138 0.133 0.128 0.132 0.98 1.44 ΔIncome 0.165 0.154 0.145 0.144 0.145 0.138 0.76 0.08 Income_decrease 0.189 0.000 0.260 0.000 0.259 0.000 –3.84*** –3.83*** Two_year_decrease 0.057 0.000 0.093 0.000 0.101 0.000 –3.30*** –3.29*** GDP_growth 10.647 10.100 10.630 10.100 10.786 10.100 –1.22 –1.47 Employee_intensity 1.657 0.994 2.239 1.508 3.079 1.886 –7.60*** –10.87*** Fixed_asset_intensity 2.261 1.605 2.503 1.670 2.877 1.893 –3.15*** –4.37*** Leverage 0.471 0.478 0.517 0.531 0.503 0.512 –4.74*** –4.87*** Marketisation_index 8.826 8.960 8.580 8.660 7.703 7.420 8.74*** 8.88*** Independent_director 0.341 0.333 0.355 0.333 0.342 0.333 –1.88* –1.24 Duality 0.057 0.000 0.103 0.000 0.096 0.000 –3.29*** –3.29*** Managerial_share 0.001 0.000 0.006 0.000 0.004 0.000 –3.39*** –0.75 Note: For deﬁnitions of the variables, please refer to Table 2. ***, **, * indicate signiﬁcance levels of 1%, 5% and 10% respectively Wang (2006), Yu (2007), and others. A total of 33.5% of observations are audited by Big Ten Domestic (mean of Domestic_big10). All the statistics of the control vari- ables are within reasonable ranges. Panel B reports the descriptive statistics for subs- amples partitioned by the types of audit ﬁrms. The results show that the overall condition is better in listed companies audited by Big Four International ﬁrms. Finan- cial leverage (Leverage), of the likelihood of having a decreased income, the likeli- hood of two consecutive years of decreased income, and the likelihood of CEO duality are lower for such ﬁrms and, on average, the local marketisation level is higher. These results are consistent with previous studies (e.g. Qi et al., 2004; Wang & Wang, 2006). Table 4 shows the correlation coefﬁcients of the main variables. The correlation matrix shows no strong correlation between the independent variables and the control variables. 304 Liang et al. Table 4. Correlation coefﬁcients. AB C D EF G H I J K L M (A) ΔCost 1 *** (B) ΔIncome 0.294 1 *** *** (C) Income_decrease –0.200 –0.536 1 *** (D) Big4 0.007 0.010 –0.037 1 *** (E) Domestic_big10 0.009 –0.001 0.008 –0.174 1 *** *** *** *** (F) Two_year_decrease –0.142 –0.344 0.556 –0.032 –0.008 1 *** *** *** *** *** (G) GDP_growth –0.073 0.045 –0.088 –0.012 –0.042 –0.040 1 *** *** *** *** *** *** *** (H) Employee_intensity –0.042 –0.246 0.178 –0.074 –0.098 0.220 –0.057 1 *** *** *** ** *** *** *** *** (I) Fixed_asset_intensity –0.058 –0.353 0.244 –0.031 –0.042 0.267 –0.042 0.410 1 *** *** *** *** *** ** (J) Leverage 0.048 0.047 0.005 –0.046 0.043 –0.005 0.043 –0.026 –0.006 1 *** ** *** *** *** *** *** (K) Marketisation_index –0.047 –0.019 0.029 0.085 0.168 0.003 0.124 –0.261 –0.066 –0.012 1 ** *** * *** *** ** * *** (L) Independent_director –0.031 0.016 –0.005 –0.018 0.087 –0.025 0.117 –0.129 –0.027 0.021 0.223 1 ** ** *** * (M) Duality 0.005 –0.000 –0.009 –0.032 0.017 –0.005 0.001 0.011 0.007 –0.031 0.038 0.020 1 ** ** *** *** *** * *** *** *** *** *** (N) Managerial_share 0.026 0.012 –0.031 –0.033 0.036 –0.033 –0.015 –0.024 –0.034 –0.081 0.144 0.071 0.161 Note: For deﬁnitions of the variables, please refer to Table 2. ***, **, * indicate signiﬁcance levels of 1%, 5% and 10% respectively. China Journal of Accounting Studies 305 4. Empirical analysis 4.1. Testing H1: types of external auditors and cost stickiness Table 5 shows the regression results of model (1). Column (1) contains only ΔIncome and ΔIncome*Income_decrease, while columns (2) to (6) further include control vari- ables and our variables of interest. The regression result shows that the F-values of the regression model are mostly above 20 and the adjusted R values are about 0.10, close to the results of Sun and Liu (2004). The coefﬁcient of ΔIncome*Income_decrease is negative and signiﬁcant at the 1% level in column (1), suggesting that cost stickiness exists in China’s listed companies, which is consistent with the results of Gong et al. (2010) and Sun and Liu (2004). The sign and signiﬁcance of the coefﬁcient of ΔIncome*Income_decrease remain unchanged in column (2) when control variables are included. Columns (3) to (5) present the regression results for the subsamples by type of external auditors. The coefﬁcient of ΔIncome*Income_decrease is positive and not signiﬁcant in column (3) for companies audited by Big Four International ﬁrms, nega- tive but not signiﬁcant in column (4) for companies audited by Big Ten Domestic ﬁrms, and negative and signiﬁcant at the 5% level in column (5) for companies audited by other audit ﬁrms. Columns (6) and (7) include the interaction term of three variables in the model to test for H1a and H1b. The coefﬁcient of ΔIncome*Income_ decrease*Big4 is positive and signiﬁcant at the 10% level in column (6), meaning that Big Four International audited ﬁrms have signiﬁcantly less cost stickiness compared with other ﬁrms. Column (7) further includes the interaction of Domebig10 and ΔIncome*Income_decrease. The coefﬁcient of ΔIncome*Income_decrease*Big4 remains positive and signiﬁcant at the 10% level, but the coefﬁcient of ΔIncome*Inco- me_decrease*Big10 is positive but not signiﬁcant, which means that there is no signiﬁ- cant difference in cost stickiness between ﬁrms audited by Big Ten Domestic and by the non-Big Ten Domestic. The results in columns (3) to (7) support H1a but not H1b. We think that the non-signiﬁcant result for H1b is not surprising. For different reasons, such as a less diversiﬁed business, brain drain, and government intervention, most pre- vious studies have failed to ﬁnd signiﬁcant audit quality differences between domestic audit ﬁrms of different sizes (Fang, Hong, & Li, 2004;Wu, 2006; Yuan & Li, 2003). In addition (untabled), the joint test shows that the sum of the coefﬁcients of ΔIncome*Income_decrease and ΔIncome*Income_decrease*Big4 is not signiﬁcantly different from zero. 4.2. Testing H2: types of external auditors, agency costs, and cost stickiness For H2, we separate the entire sample into two subsamples according to the proportion of shares held by the largest shareholders. If the proportion is above the median, the observation is classiﬁed as a high stake of major shareholders; otherwise the observa- tion is classiﬁed as a low stake of major shareholders. We want to use subsample regressions to explore the difference in the restrictive roles of Big Four International and Big Ten Domestic audit ﬁrms on cost stickiness in different environments. Table 6 shows the regression results. The coefﬁcient of ΔIncome*Income_ decrease*Big4 is not signiﬁcant in column (1) for ﬁrms with high stakes of major shareholders, suggesting that Big Four International audit ﬁrms do not signiﬁcantly reduce cost stickiness when the largest shareholder’s holding ratio is high. However, the coefﬁcient of ΔIncome*Income_decrease*Big4 is positive and signiﬁcant at the 1% level in column (2) for ﬁrms with low stakes of major shareholders, which means that 306 Liang et al. Table 5. Types of external auditors and the cost stickiness. (1) Full (2) Full (3) The Big (4) The Big Ten (5) Others (6) Full (7) Full sample sample Four Domestic domestic sample sample 0.309*** 0.294*** 0.351*** 0.332*** 0.274*** 0.294*** 0.294*** (10.68) (10.10) (3.51) (6.69) (7.40) (10.10) (10.08) ΔIncome*Income_decrease –0.151*** –0.145** 0.076 –0.116 –0.153** –0.148*** –0.169*** (–3.01) (–2.57) (0.48) (–1.45) (–2.14) (–2.61) (–2.71) ΔIncome*Income_decrease*Big4 0.214* 0.238* (1.70) (1.84) 0.024** 0.031*** (2.20) (2.71) ΔIncome*Income_decrease 0.086 *Domestic_big10 (1.27) Domestic_big10 0.020*** (2.68) Two_year_decrease –0.085*** –0.041 –0.078*** –0.086*** –0.085*** –0.083*** (–4.86) (–0.57) (–3.08) (–3.77) (–4.83) (–4.82) GDP_growth –0.022*** 0.014 –0.022*** –0.025*** –0.022*** –0.022*** (–5.43) (1.26) (–2.65) (–4.95) (–5.35) (–5.40) Employee_intensity 0.001 –0.008 0.001 0.001 0.001 0.001 (0.39) (–1.50) (0.34) (0.28) (0.42) (0.38) Fixed_asset_intensity 0.002 0.017*** 0.003 0.002 0.002 0.002 (1.04) (2.84) (1.03) (0.52) (1.03) (1.08) Leverage 0.075*** 0.114 0.046 0.082*** 0.076*** 0.074*** (3.81) (1.56) (1.55) (2.86) (3.85) (3.73) Marketisation_index –0.005*** –0.001 –0.007*** –0.005** –0.005*** –0.005*** (–3.09) (–0.19) (–2.89) (–2.13) (–3.20) (–3.39) Independent_director –0.099* –0.159 –0.108 –0.095 –0.099* –0.100* (–1.70) (–1.12) (–1.01) (–1.21) (–1.70) (–1.72) Duality 0.001 0.044* –0.004 0.003 0.001 0.001 (0.12) (1.83) (–0.26) (0.19) (0.14) (0.13) Managerial_share 0.436*** 1.131*** 0.497*** 0.417*** 0.442*** 0.445*** (5.07) (3.61) (4.18) (3.59) (5.12) (5.22) China Journal of Accounting Studies 307 Constant 0.139*** 0.137*** 0.057 0.165*** 0.140*** 0.137*** 0.134*** (8.74) (4.93) (0.84) (3.31) (3.71) (4.90) (4.82) Year Yes Yes Yes Yes Yes Yes Yes Observations 10,494 10,494 593 3,520 6,381 10,494 10,494 Adjusted R 0.101 0.107 0.130 0.157 0.088 0.107 0.108 F value (model) 34.15 27.62 7.737 14.67 14.71 26.14 25.08 F value (ΔIncome* Income_decrease) (3) vs (5)=1.48 Note: For deﬁnitions of the variables, please refer to Table 2. ***, **, * indicate signiﬁcance levels of 1%, 5% and 10% respectively. 308 Liang et al. Table 6. Types of external auditors, agency costs and the expense stickiness. (1) High stake (2) Low stake (3) High (4) Low of major of major level of level of shareholders shareholders marketisation marketisation ΔIncome 0.332*** 0.241*** 0.271*** 0.317*** (7.67) (5.88) (7.37) (6.98) ΔIncome*Income_decrease –0.161 –0.137* –0.109 –0.217** (–1.27) (–1.83) (–1.38) (–2.38) ΔIncome*Income_decrease*Big4 0.011 0.690*** 0.079 0.717** (0.09) (2.98) (0.77) (2.36) Big4 0.011 0.064*** 0.030** 0.028 (0.77) (3.43) (2.49) (1.14) ΔIncome*Income_decrease 0.172 0.059 0.035 0.150 *Domestic_big10 (1.47) (0.71) (0.40) (1.20) Domestic_big10 0.018* 0.021* 0.010 0.032*** (1.77) (1.81) (1.04) (2.71) Two_year_decrease –0.048 –0.103*** –0.084*** –0.077*** (–1.54) (–4.64) (–3.97) (–2.84) GDP_growth –0.017*** –0.027*** –0.014** –0.027*** (–3.21) (–3.89) (–2.07) (–4.53) Employee_intensity –0.001 0.002 0.000 0.001 (–0.50) (0.65) (0.05) (0.34) Fixed_asset_intensity 0.004 0.002 0.003 0.002 (1.09) (0.72) (1.02) (0.54) Leverage 0.040 0.101*** 0.045* 0.103*** (1.42) (3.51) (1.82) (3.21) Marketisation_index –0.009*** –0.002 –0.010*** –0.005 (–4.09) (–0.93) (–2.75) (–1.04) Independent_director –0.102 –0.110 –0.068 –0.125 (–1.15) (–1.28) (–0.95) (–1.36) Duality 0.011 –0.004 0.008 –0.006 (0.68) (–0.30) (0.54) (–0.33) Managerial_share 0.490*** 0.510*** 0.516*** 0.151 (2.89) (5.16) (6.70) (0.68) Constant 0.174*** 0.114** 0.180*** 0.123** (4.38) (2.56) (3.59) (2.52) Year Yes Yes Yes Yes Observations 5,249 5,245 5,263 5,231 Adjusted R 0.150 0.078 0.126 0.095 F value (model) 15.24 12.79 18.28 11.89 F value (ΔIncome* (1) vs (2) = 6.46** (3) vs (4) = 4.01** Income_decrease) Note: For deﬁnitions of the variables, please refer to Table 2. ***, **, * indicate signiﬁcance levels of 1%, 5% and 10% respectively. Big Four International audit ﬁrms can reduce cost stickiness signiﬁcantly when the largest shareholder’s holding ratio is low. In addition, the F-value of the Chow test of the coefﬁcients of ΔIncome*Income_decrease*Big4 between the two groups is 6.46 and signiﬁcant at the 1% level. All of these results show that Big Four International audit ﬁrms play an active role when major shareholders hold fewer stakes and when agency costs are high. Similarly, the coefﬁcient of ΔIncome*Income_decrease*Big4 is not signiﬁcant in column (3) for ﬁrms in areas with higher levels of marketisation, indi- cating that Big Four International audit ﬁrms do not suppress cost stickiness signiﬁ- China Journal of Accounting Studies 309 cantly when ﬁrms are located in areas with well-developed markets. On the contrary, the coefﬁcient of ΔIncome*Income_decrease*Big4 is positive and signiﬁcant at the 5% level in column (4) for ﬁrms in areas with lower levels of marketisation, which means that Big Four International ﬁrms can signiﬁcantly restrict cost stickiness when ﬁrms are located in areas with less developed markets. In addition, the F-value of the Chow test of the coefﬁcients of ΔIncome*Income_decrease*Big4 is 4.01 and signiﬁcant at the 5% level. All these results suggest that Big Four International audit ﬁrms play a positive role in reducing cost stickiness when the market is not well developed and when agency cost is high. We do not ﬁnd evidence that the restrictive roles on cost stickiness are signiﬁcantly different between Big Ten Domestic and non–Big Ten Domestic audit ﬁrms. The above results together show in the role of Big Four International ﬁrms in mitigating agency costs and reducing cost stickiness changes under different circum- stances. 4.3. Robustness tests 4.3.1. Test using propensity score matching (PSM) So far, we have not ruled out the impact of self-selection on our results. For instance, companies with strict internal control may tend to choose audit services provided by Big Four International audit ﬁrms, whose services are of higher quality (e.g. Thornton & Moore, 1993). At the same time, these companies may exhibit lower cost stickiness due to strict internal control (Chen, Lu, & Sougiannis, 2012). If this is the case, our result that companies audited by Big Four International have lower cost stickiness is just a natural result of the companies’ choices. To address such concerns, we use the PSM method to further test our results. The PSM method can eliminate the effect of self-selection by matching the treated group (Big Four International-audited ﬁrms in this article) with the control group (non-Big Four International-audited ﬁrms) and is widely employed in related accounting and auditing research (e.g. Carcello, Vanstraelen, & Willenborg, 2009; Lawrence, Minutti-Meza, & Zhang, 2011). Speciﬁcally, the PSM test contains two stages. The ﬁrst stage is to grade based on choice tendency and the second stage is to match samples based on the grades gener- ated. The grading is based on a logit model whose dependent variable is an indicator variable of Big Four International audited ﬁrms (Big4). We choose independent vari- ables and control variables according to Wang et al. (2008). The grading model is Big4 ¼ a þ a SOE þ a Growth þ a Size þ a CROA þ a Leverage þ a CurrentR 0 1 2 3 4 5 6 þ a InventoryR þ a RecR þ RYear þ e (2) 7 8 where SOE is a dummy variable that equals one for state-owned companies and zero otherwise, Growth is the growth rate of operating revenue, Size is the natural logarithm of total assets, CROA is the proﬁt from operations divided by total assets, Lev is ﬁnan- cial leverage, CurrentR is the ratio of current assets to current liability, InventoryR is the ratio of inventory to total assets, and RecR is the ratio of accounts receivable to total assets. We also control for year ﬁxed effects. The ﬁrst-stage regression test shows that Size, Leverage, and other factors have signiﬁcant inﬂuence on whether a company hires a Big Four International audit ﬁrm. Matching by grades, we generate a total sam- ple of 1,084 observations, including both the treated and control groups, and we use this sample to conduct the second-stage regression. 310 Liang et al. Table 7 shows the regression results. Column (1) is based on full sample regression, while columns (2) to (5) report the results of subsamples grouped by largest share- holder’s holding ratio (columns (2) and (3)) or local marketisation levels (columns (3) and (4)). The results shows that the coefﬁcient of ΔIncome*Income_decrease*Big4 is positive and signiﬁcant at the 10% level in column (1), suggesting the restrictive role of Big Four International audit ﬁrms in cost stickiness. The coefﬁcient of ΔIncome*In- come_decrease*Big4 is not signiﬁcant in column (2) for ﬁrms with a high stake of major shareholders, but is positive and signiﬁcant at the 1% level in column (3) for ﬁrms with a low stake of major shareholders. Meanwhile, the F-value of the Chow test of the coefﬁcients of ΔIncome*Income_decrease*Big4 in the two columns is 3.86 and signiﬁcant at the 5% level. The coefﬁcient of ΔIncome*Income_decrease*Big4 is not signiﬁcant in column (4) for ﬁrms in areas with a well-developed market and is positive and signiﬁcant at the 10% level in column (5) for ﬁrms in areas with lower marketisa- tion levels. In addition, the F-value of the Chow test of the two coefﬁcients is 4.17 and signiﬁcant at the 5% level. These results are qualitatively the same as the results in pre- vious sections. The results in Table 7 show that using PSM does not change the out- come we found earlier. 4.3.2. Test on industrial listed companies Weiss (2010) uses only industrial listed companies to examine cost stickiness for the following two reasons: ﬁrst, the cost and proﬁt model of companies from a single industry allows for greater comparability. Second, industrial listed companies face more severe competition and using such a sample can better avoid the inﬂuence of monopo- listic prices. Therefore, we choose a subsample of industrial ﬁrms according to the industry classiﬁcation of the China Securities Market and Accounting Research (CSMAR) and conduct robust tests. Table 8 shows the regression results. Column (1) presents the full sample regres- sion, while columns (2) to (5) report the results of subsamples grouped by the largest shareholder’s holding ratio (columns (2) and (3)) or local marketisation levels (columns (3) and (4)). The results shows that the coefﬁcients of ΔIncome*Income_decrease*Big4 are positive and signiﬁcant at the 5% level in columns (1), (3), and (5), but are not sig- niﬁcant in column (2) for ﬁrms with a high stake of major shareholders or in column (4) for ﬁrms in areas with lower marketisation levels. These results support the positive role of the Big Four International audit ﬁrms in reducing cost stickiness and also the inference that the restrictive role varies with the largest shareholder’s holding ratio and local marketisation level. The results in Table 8 show that using a subsample of indus- trial ﬁrms does not change the outcome we found earlier. 4.3.3. Control for discretionary accruals Chen, Song, and Shi (2012) ﬁnd that companies may use a big bath to manage earnings to avoid future losses and such behaviour can lead to the overestimation of stickiness. Thus, we follow Dechow, Sloan, and Sweeney (1995) and use the adjusted Jones model to calculate discretionary accruals (DA). We then include DA and an interaction term ΔIncome*Income_decrease*DA in our model and re-estimate the model. The results are presented in Table 9. Column (1) presents the full sample regression, while columns (2) to (5) report the results of subsamples grouped by the largest shareholder’s holding ratio (columns (2) and (3)) or local marketisation levels (columns (3) and (4)). The China Journal of Accounting Studies 311 Table 7. Test using the PSM method. (1) Full (2) High stake of major (3) Low stake of major (4) High level of (5) Low level of sample shareholders shareholders marketisation marketisation ΔIncome 0.469*** 0.479*** 0.485*** 0.453*** 0.487*** (8.94) (5.27) (11.54) (7.62) (4.35) ΔIncome*Income_decrease –0.431*** –0.420** –0.629*** –0.163 –0.531** (–3.21) (–2.44) (–3.18) (–0.88) (–2.45) ΔIncome*Income_decrease*Big4 0.306* 0.144 0.753*** –0.035 0.848* (1.81) (0.92) (2.87) (–0.19) (1.96) Big4 –0.014 –0.038* 0.047 –0.010 –0.018 (–0.81) (–1.84) (1.63) (–0.51) (–0.56) Two_year_decrease 0.028 0.130* –0.181** 0.006 0.087 (0.50) (1.75) (–2.57) (0.12) (0.64) GDP_growth 0.001 –0.006 0.020 0.016 –0.015 (0.10) (–0.56) (1.42) (1.29) (–0.99) Employee_intensity –0.005 –0.004 –0.005 –0.001 –0.005 (–0.99) (–0.71) (–0.90) (–0.13) (–0.77) Fixed_asset_intensity 0.005 0.005 0.009 0.012** –0.001 (1.04) (0.92) (0.89) (2.05) (–0.12) Leverage 0.010 –0.003 –0.006 0.060 –0.017 (0.16) (–0.04) (–0.06) (0.89) (–0.13) Marketisation_index –0.006 –0.010* 0.004 –0.016 0.000 (–1.25) (–1.66) (0.51) (–1.63) (0.00) Independent_director –0.139 –0.015 –0.473** –0.160 –0.090 (–1.02) (–0.09) (–2.08) (–0.90) (–0.41) Duality 0.065* 0.051 0.103** 0.060 0.068 (1.77) (1.04) (2.03) (1.19) (1.60) Managerial_share –0.539 –3.056*** –0.391 –0.414 –0.817*** (–1.27) (–3.90) (–0.95) (–0.58) (–2.65) Constant 0.140** 0.158** 0.081 0.171 0.125 (2.22) (2.05) (0.73) (1.56) (0.98) Year Yes Yes Yes Yes Yes Observations 1,084 725 359 641 443 Adjusted R 0.179 0.143 0.298 0.254 0.101 F value (model) 6.908 4.905 13.89 8.414 45.14 F value (ΔIncome* (2) vs (3) = 3.86** (4) vs (5) = 4.17** Income_decrease) Note: For deﬁnitions of the variables, please refer to Table 2. ***, **, * indicate signiﬁcance levels of 1%, 5% and 10% respectively. 312 Liang et al. Table 8. Test on industrial listed companies. (1) Full (2) High stake of major (3) Low stake of major (4) High level of (5) Low level of sample shareholders shareholders marketisation marketisation ΔIncome 0.328*** 0.334*** 0.305*** 0.324*** 0.329*** (6.82) (5.54) (3.56) (5.13) (4.83) ΔIncome*Income_decrease –0.121 –0.007 –0.122 –0.140 –0.107 (–1.22) (–0.06) (–0.91) (–1.17) (–0.73) ΔIncome*Income_decrease*Big4 0.561** 0.301 0.700** 0.387 0.750** (2.47) (1.04) (2.33) (1.62) (2.48) Big4 0.018 –0.006 0.042** 0.026* 0.001 (1.26) (–0.35) (2.06) (1.73) (0.05) ΔIncome*Income_decrease 0.052 –0.011 0.063 0.054 0.041 *Domestic_big10 (0.47) (–0.09) (0.45) (0.42) (0.19) Domestic_big10 0.017* 0.006 0.024* 0.015 0.021 (1.93) (0.54) (1.75) (1.31) (1.51) Two_year_decrease –0.062*** –0.020 –0.083*** –0.070** –0.054 (–2.81) (–0.57) (–2.85) (–2.40) (–1.60) GDP_growth –0.023*** –0.021*** –0.024** –0.016* –0.028*** (–4.53) (–2.94) (–2.57) (–1.77) (–4.01) Employee_intensity –0.001 –0.004 0.001 –0.001 –0.002 (–0.34) (–1.05) (0.29) (–0.13) (–0.60) Fixed_asset_intensity 0.007 0.008 0.007 0.003 0.010 (1.41) (1.43) (1.07) (0.69) (1.28) Leverage 0.057** 0.019 0.089** 0.024 0.081** (2.32) (0.57) (2.40) (0.72) (2.19) Marketisation_index –0.004** –0.008*** –0.001 –0.013*** 0.000 (–1.97) (–2.80) (–0.18) (–2.83) (0.05) Independent_director –0.063 –0.089 –0.051 –0.100 –0.028 (–0.78) (–0.72) (–0.47) (–0.97) (–0.23) Duality –0.010 –0.001 –0.015 –0.003 –0.017 (–0.66) (–0.04) (–0.75) (–0.14) (–0.79) Managerial_share 0.329*** 0.305 0.382*** 0.441*** –0.060 (3.30) (1.43) (3.29) (4.65) (–0.38) China Journal of Accounting Studies 313 Constant 0.132*** 0.198*** 0.072 0.239*** 0.082 (3.46) (3.93) (1.18) (3.50) (1.40) Year Yes Yes Yes Yes Yes Observations 6,236 3,295 2,941 2,830 3,406 Adjusted R 0.106 0.128 0.086 0.138 0.089 F value (model) 14.33 8.998 7.523 11.65 7.891 F value (ΔIncome* (2) vs (3) = 0.88 (4) vs (5) = 1.00 Income_decrease) Note: For deﬁnitions of the variables, please refer to Table 2. ***, **, * indicate signiﬁcance levels of 1%, 5% and 10% respectively. 314 Liang et al. Table 9. Control for discretionary accruals. (1) Full (2) High stake of major (3) Low stake of major (4) High level of (5) Low level of sample shareholders shareholders marketisation marketisation ΔIncome 0.304*** 0.358*** 0.240*** 0.270*** 0.348*** (10.63) (8.56) (5.85) (7.36) (7.99) ΔIncome*Income_decrease –0.170*** –0.177 –0.130** –0.104 –0.237*** (–3.00) (–1.42) (–1.97) (–1.36) (–3.11) ΔIncome*Income_decrease*DA 0.773*** 0.488 0.846*** 0.350 1.322*** (3.84) (1.47) (3.37) (1.58) (3.84) DA –0.021 –0.091 0.016 0.019 –0.091 (–0.38) (–1.27) (0.18) (0.31) (–0.97) ΔIncome*Income_decrease*Big4 0.207 –0.020 0.645*** 0.070 0.651** (1.63) (–0.16) (2.96) (0.68) (2.44) Big4 0.029** 0.009 0.062*** 0.029** 0.022 (2.56) (0.62) (3.38) (2.44) (0.92) ΔIncome*Income_decrease 0.063 0.105 0.066 0.030 0.077 *Domestic_big10 (0.98) (0.85) (0.82) (0.36) (0.63) Domestic_big10 0.020*** 0.016 0.022* 0.010 0.029** (2.62) (1.58) (1.89) (1.06) (2.43) Two_year_decrease –0.085*** –0.050 –0.107*** –0.087*** –0.076*** (–5.01) (–1.62) (–4.80) (–4.06) (–2.86) GDP_growth –0.022*** –0.017*** –0.026*** –0.014** –0.027*** (–5.32) (–3.28) (–3.81) (–2.06) (–4.58) Employee_intensity –0.001 –0.002 0.000 –0.000 –0.001 (–0.31) (–0.90) (0.18) (–0.13) (–0.49) Fixed_asset_intensity 0.002 0.004 0.002 0.003 0.000 (1.12) (1.11) (0.80) (1.17) (0.09) Leverage 0.058*** 0.022 0.086*** 0.042* 0.065** (2.93) (0.77) (3.07) (1.70) (1.99) Marketisation_index –0.005*** –0.009*** –0.002 –0.010*** –0.005 (–3.58) (–4.16) (–1.05) (–2.78) (–1.06) Independent_director –0.098* –0.089 –0.119 –0.068 –0.120 (–1.70) (–1.00) (–1.43) (–0.95) (–1.34) China Journal of Accounting Studies 315 Duality 0.003 0.011 –0.002 0.008 –0.000 (0.27) (0.65) (–0.12) (0.55) (–0.02) Managerial_share 0.449*** 0.530*** 0.497*** 0.517*** 0.137 (5.27) (3.26) (4.95) (6.77) (0.60) Constant 0.143*** 0.175*** 0.129*** 0.183*** 0.141*** (5.29) (4.57) (2.99) (3.63) (3.07) Year Yes Yes Yes Yes Yes Observations 10,481 5,244 5,237 5,262 5,219 Adjusted R 0.117 0.162 0.086 0.128 0.114 F value (model) 25.48 15.74 13.02 17.16 13.65 F value (ΔIncome* (2) vs (3) = 6.71*** (4) vs (5) = 4.28** Income_decrease) Note: DA is discretionary accruals, which is calculated following Dechow et al. (1995). The other variables are deﬁned in Table 2. ***, **, * indicate signiﬁcance levels of 1%, 5% and 10% respectively. 316 Liang et al. results show that, after Da and the interaction term ΔIncome*Income_decrease*Da are controlled for, the coefﬁcient of ΔIncome*Income_decrease*Big4 is positive and signiﬁ- cant in column (1), while the coefﬁcient of ΔIncome*Income_decrease*Da is positive and signiﬁcant at the 1% level. The combined results indicate that Big Four International audit ﬁrms reduce cost stickiness mainly by inﬂuencing managers’ earnings management behaviour, since the introduction of discretionary accruals lowers the explanatory power of Big Four International audit ﬁrms for cost stickiness. Further subsample tests suggest that the coefﬁcients of ΔIncome*Income_decrease*Big4 are positive and signiﬁcant at the 1% and 5% levels, respectively, in columns (3) and (5), but are not signiﬁcant in col- umns (2) and (4). The results of Table 9 suggest that controlling discretionary accruals will not change our earlier outcomes. 4.3.4. Other robustness tests In addition, we carry out the following unreported tests. (1) Following Anderson et al. (2003), Chen, Lu, and Sougiannis (2012), and Sun and Liu (2004), we incorpo- rate the interactions of the four economic factors Two_year_decrease, GDP_growth, Employee_intensity, and Fixed_asset_intensity and stickiness (ΔIncome*Income_de- crease) or the interaction terms of all control variables and stickiness (ΔIncome*Inco- me_decrease) into the model. The results are qualitatively the same for both methods, with coefﬁcients of ΔIncome*Income_decrease*Big4 that are positive but not signiﬁ- cant in the full sample regressions and a signiﬁcant difference in the coefﬁcients of ΔIncome*Income_decrease*Big4 between different subsamples. (2) Following Chen, Song, and Shi (2012), we measure ﬁxed capital intensity by the ratio of ﬁxed assets to operating income. (3) We substitute Da with its absolute value. (4) We use yearly total income ranking to identify the Big Ten Domestic audit ﬁrms. (5) We choose the Big Five Domestic or the Big 20 Domestic instead of the Big Ten Domestic. None of the robustness tests changes the previous results qualitatively. 5. Conclusions and limitations Cost stickiness is an important problem for the issue of cost behaviour and an impor- tant measure of enterprise production and managerial efﬁciency. However, for a long time, related studies on cost stickiness have been limited and mainly focused on the impact of internal corporate governance mechanisms (Calleja et al., 2006; Chen, Lu, & Sougiannis, 2012). We introduce an important external governance mechanism, namely external audit, into this topic and propose that an external audit, which can mitigate information asymmetry and reduce internal agency costs through independent audit work, can play a restrictive role in a ﬁrm’s cost stickiness. Using China A-shares listed ﬁrms from 2002 to 2010, we test this hypothesis empirically. The results show that cost stickiness is less pronounced for ﬁrms audited by the Big Four International audit ﬁrms than for ﬁrms audited by non-Big Four Inter- national audit ﬁrms, but such a difference in cost stickiness is not found between ﬁrms audited by the Big Ten Domestic audit ﬁrms and by non-Big Ten Domestic audit ﬁrms. In addition, we ﬁnd that the restrictive role of the Big Four International is more pro- nounced when a ﬁrm’s largest shareholder’s holding ratio is lower or when a ﬁrm is located in an area with a lower marketisation level. The results are unchanged when we use PSM or industrial listed company subsamples or when we control for discre- tionary accruals. Our study enriches the literature on factors inﬂuencing cost stickiness China Journal of Accounting Studies 317 and indicates new incentives for the demand for high-quality audit services and implies the higher audit quality of Big Four International audit ﬁrms. Our paper also has the following limitations. First, there can be model misspeciﬁca- tion. (1) Our basic model follows Anderson et al. (2003), Chen, Lu, and Sougiannis (2012) and Sun and Liu (2004), but the missing quadratic term in the model can lead to econometric errors. (2) The selection of economic factors can cause problems. There are no conclusive standards for this issue and the domestic literature discusses this even less. This paper controls for the four most frequently selected factors, which are the indi- cator of two consecutive years of a decline in operating income, economic growth, human capital density, and ﬁxed capital intensity. However, this does not mean that the inﬂuence from the contract perspective and efﬁciency perspective is perfectly controlled for in our model. More variables reﬂecting these two perspectives may be gradually identiﬁed in the future. (3) There are other measures of cost stickiness. Weiss (2010) measures stickiness using quarterly operating ﬂuctuations, but seasonal sales will natu- rally affect the estimation of stickiness (e.g. the seasonal sales ﬂuctuation of electric appliances may not suggest a decline of sales). In addition, quarterly ﬁnancial statements are not mandatorily audited and may suffer more from earnings management. As a result, we do not use this method for robustness checks. Second, the self-selection prob- lem may still affect our results. Although we use PSM to match ﬁrms audited by Big Four International ﬁrms with control ﬁrms, we have to admit that such methods can only mitigate instead of rule out the impact of the self-selection problem and we should inter- pret the results with caution. These limitations can be addressed in future research. Acknowledgements This research is supported by grants from the National Natural Science Foundation of China (No. 71402198), the Beijing Municipal Commission of Education ‘Joint Construction Project’, the Beijing Municipal Commission of Education ‘Pilot Reform of Accounting Discipline Clustering’, ‘2011 Synergetic Innovation’ Key Project on ‘Development of Public Accounting Profession’ and ‘121 Talent Project for Young Doctoral Graduates Development 2014’ of the Central University of Finance and Economics. Notes 1. Chen, Lu, and Sougiannis (2012) provide direct evidence for this argument. They ﬁnd that cost stickiness has a positive correlation with corporate free cash ﬂow and this positive cor- relation can be weakened when the corporate governance level is higher. Calleja et al. (2006) ﬁnd that expense stickiness in France and Germany is higher than in the United States or Britain in their international study and they conjecture that the difference in corpo- rate governance is the main reason. 2. China is a large country where the levels of market development and government interven- tion vary signiﬁcantly across the 30 areas (provinces, autonomous regions or municipalities) (Jian & Wong, 2010). Fan, Wang, and Zhu (2010) offer an index that measures the develop- ment of marketisation of different areas in China. 3. In different years, it can be the Big Five International or the Big Eight International instead of Big Four International. For the sake of brevity, we use the term Big Four International for all conditions without differentiation. All these Big N international audit ﬁrms refer to the well-recognised top international audit ﬁrms. 4. The ﬁndings of these studies conﬂict for the following reasons, in addition to the difference in research focus (discretionary accruals, audit opinion types, ERC (earnings response coef- ﬁcient), and conservatism): ﬁrst, the sample period is short. Most sample periods are one year to three years and the longest are ﬁve years. The shorter the sample period, the more the variables of interest can ﬂuctuate, especially discretionary accruals. Second, the differ- 318 Liang et al. ence in the sample size of ﬁrms audited by Big Four International and non-Big Four Inter- national is not well addressed. Third, the self-selection problem is paid little attention. Our paper, instead, chooses a nine-year sample period (from 2002 to 2010) to conduct the empirical tests and methods such as propensity score matching (PSM) to mitigate the poten- tial problem. 5. The SME started in 2004 and GEM started in 2009. But the government stopped the IPO approval during the non-tradable share reform (2005–2006), so companies listed before 2007 are mainly on the main board. 6. Other reasons include the consideration of cost adjustment and managers’ optimism (Banker et al., 2011). Sun and Liu (2004) attribute these reasons to an opportunistic behaviour per- spective, a contract perspective, and an efﬁciency perspective. 7. However, their research design does not control for common variables, economic factor variables, or year or industry ﬁxed effects. 8. Calleja et al. (2006) mention external supervision theoretically, but do not conduct empirical research. 9. Such as overextending the number of employees (Williamson, 1963) and building luxurious ofﬁces. 10. Controlling earnings management is not the only way external governance works. If a com- pany’s performance worsens and managers still spend 1 million RMB to build luxurious ofﬁces, they can cover the adverse inﬂuence of such expenses on performance through earn- ings management. A capable and independent auditor can either correct the earnings man- agement during the auditing procedure, which reduces the degree of earnings management, or alert shareholders by signing non-standard audit opinions without correcting for earnings managements (in such cases, the degree of earnings management is unchanged). 11. Although many variables have been used in prior studies to measure agency costs, such as the administrative expense rate and free cash ﬂow, most of these variables might present the consequence of the whole corporate governance including auditing (Gul & Tsui, 1998; Li, 2007). 12. Our sample starts in 2002 because the concept of the Big Four International tends to have become stable with the voluntary surrender by Arthur Andersen of its licence to practice in the US. Also, the Chinese CPA Association started issuing its Comprehensive Evaluation of the Top 100 Accounting Firms in 2002, which provides a good reference for deﬁning the Big Ten Domestic audit ﬁrms. 13. The ﬁrm ranking is derived from yearly Comprehensive Evaluation of the Top 100 Account- ing Firms released by China CPA Association (http://www.cicpa.org.cn/Column/swszhpm). 14. These differences further support the necessity of using PSM as a robustness test. 15. Some later tests show that the sum of the coefﬁcients of ΔIncome*Income_decrease and ΔIncome*Income_decrease*Big4 is signiﬁcantly positive, meaning that there is an anti- stickiness phenomenon. This kind of result is mainly found in groups with lower largest shareholder’s shareholding ratio and groups with a lower market index. We conjecture that one possible reason is that the agency cost in this kind of company is higher. When sales decrease, managers trying to achieve performance goals may have to adopt real earnings management if their audit ﬁrms (Big Four International) effectively restrict them from con- ducting earnings management. In such a situation, excessive cost cutting can lead to an anti-stickiness phenomenon. Cohen and Zarowin (2010) and Zang (2012) both point out that the degree of real earnings management is higher when ﬁrms are audited by the Big Eight International. We do not ﬁnd an anti-stickiness phenomenon when PSM methods are used. 16. The medians of the largest shareholder’s holding ratio (Sh) and the market index (Marketi- sation_index) are 0.357 and 7.92, respectively. 17. We also estimate the model using 2×2 subsamples based on whether the ﬁrm is audited by Big Four International ﬁrms and whether the largest shareholder’s holding ratio (or market index) is higher than the median. The results show that the cost stickiness in groups when ﬁrms are audited by non-Big Four International ﬁrms and when the largest shareholder’s holding ratio (or when the market index is lower than the median) is the strongest. 18. The result is more signiﬁcant if we compare the Big Four International sample with the non-Big Ten Domestic sample (e.g. the control group does not include the Big Ten Domestic sample). China Journal of Accounting Studies 319 19. The idea is to mitigate the inﬂuence of earnings management in the form of a big bath, instead of the inﬂuence of the absolute value of earnings management. In accordance with this idea, we control for DA instead of its absolute value. In robustness tests, we control for the absolute value of discretionary accruals (AbsDA) and the interaction term ΔIncome*Income_decrease*AbsDA. 20. We think the restrictions on cost stickiness and earnings management are two parallel con- sequences. Big Four International ﬁrms can use different methods, such as restricting earn- ings management, signing non-standard audit opinions, and providing comparable ﬁnancial information of higher quality, to help shareholders make executive compensation and turn- over decisions to reduce managers’ opportunistic behaviour. Reducing earnings management is only one of these methods and, when the latter two methods are used, the degree of earn- ings management cannot completely explain the change in cost stickiness. 21. Unreported results are available from the authors upon request. 22. To address this problem, we also conduct subsample tests, but this method is convincing only in the test of the basic hypothesis. Further tests still rely on the basic model and may still be affected by potential model misspeciﬁcation (the comparison between subsamples with different ownership structure or different external environments is a Chow test of the cubic item). References Anderson, M., Banker, R., & Janakiraman, S. (2003). Are selling, general, and administrative costs ‘sticky’? Journal of Accounting Research, 41(1), 47–63. Ball, R., Robin, A., & Wu, J. S. (2003). Incentives versus standards: Properties of accounting income in four East Asian countries. Journal of Accounting and Economics, 36(1–3), 235–270. Banker, R. D., Byzalov, D., & Chen, L. T. (2013). Employment protection legislation, adjustment costs and cross-country differences in cost behavior. Journal of Accounting and Economics, 55(1), 111–127. Banker, R. D., Byzalov, D., & Plehn-Dujowich, J. M. (2011). Sticky cost behavior: Theory and evidence. Working Paper. Becker, C. L., Defond, M. L., Jiambalvo, J., & Subramanyam, K. R. (1998). The effect of audit quality on earnings management. Contemporary Accounting Research, 15(1), 1–24. Cai, C., Huang, Y. J., & Zhao, S. (2005). Empirical research on how audit quality inﬂuences earnings management: Evidence from manufacturing industry in Shanghai. Auditing Research, (2), 3–10 (in Chinese). Calleja, K., Steliaros, M., & Thomas, D. C. (2006). A note on cost stickiness: Some international comparisons. Management Accounting Research, 17(2), 127–140. Carcello, J. V., Vanstraelen, A., & Willenborg, M. (2009). Rules rather than discretion in audit standards: Going-concern opinions in Belgium. The Accounting Review, 84(5), 1395–1428. Chen, C. X., Lu, H., & Sougiannis, T. (2012). The agency problem, corporate governance, and the asymmetrical behavior of selling, general, and administrative costs. Contemporary Accounting Research, 29(1), 252–282. Chen, D. H., & Liang, S. K. (2010). Perks of managers, check-and-balance of stock ownership and its economic consequences: Empirical evidence from Chinese listed companies. Journal of Shanghai Lixin University of Commerce, (1), 19–27 (in Chinese). Chen, L., Song, L., & Shi, D. (2012). An empirical study on the upward estimation bias in cost stickiness: Evidence from Chinese listed companies. China Accounting Review, (1), 3–16 (in Chinese). China Securities Regulatory Commission and Ministry of Finance. (2001). The Regulation on the Temporary License for Foreign Accounting Firms to Conduct Business in Financial Listed Firms. (In Chinese.) Cohen, D. A., & Zarowin, P. (2010). Accrual-based and real earnings management activities around seasoned equity offerings. Journal of Accounting and Economics, 50(1), 2–19. Cohen, J., Krishnamoorthy, G., & Wright, A. M. (2002). Corporate governance and the audit pro- cess. Contemporary Accounting Research, 19(4), 573–594. DeAngelo, L. E. (1981). Auditor size and auditor quality. Journal of Accounting and Economics, 3(3), 183–199. 320 Liang et al. Dechow, P. M., Sloan, R. G., & Sweeney, A. P. (1995). Detecting earnings management. The Accounting Review, 70(2), 193–225. Defond, M. L., Wong, T. J., & Li, S. (2000). The impact of improved auditor independence on audit market concentration in China. Journal of Accounting and Economics, 28(3), 269–305. Dierynck, B., Landsman, W. R., & Renders, A. (2012). Do managerial incentives drive cost behavior? Evidence about the role of the zero earnings benchmark for labor cost behavior in private Belgian ﬁrms. The Accounting Review, 87, 1219–1246. Dopuch, N., & Simunic, D. A. (1980). The nature of competition in the auditing profession: A descriptive and normative view. In Buckley, J. and Weston, F. (Eds.), Regulation and the Accounting Profession. CA: Lifetime Learning Publications. Dou, W., Liu, X., & An, L. (2011). Ownership concentration, allocation of control rights and inefﬁcient investment in ﬁrms: Monitoring or colluding among multiple large shareholders? Journal of Management Sciences in China, 11,81–96 (in Chinese). Dye, R. (1993). Auditing standards, legal liability, and auditor wealth. Journal of Political Econ- omy, 101, 887–914. Fan, G., Wang, X. L., & Zhu, H. P. (2010). Marketization index for China’s provinces (2009), Economic Science Press (in Chinese). Fan, J., & Wong, T. J. (2005). Do external auditors perform a corporate governance role in emerging markets? Evidence from East Asia. Journal of Accounting Research, 43(1), 35–71. Fang, J. X., Hong, J. Q., & Li, R. S. (2004). Study on the inﬂuence factors of audit quality in listed companies in china: Discovery and enlightenment. Auditing Research, (6), 35–43. (in Chinese). Francis, J. R., Maydew, E. L., & Sparks, H. C. (1999). The role of big 6 auditors in the credible reporting of accruals. Auditing: A Journal of Practice & Theory, 18(2), 17–34. Gao, Q., & Wu, L. N. (2007). Large shareholder expropriation of funds and the auditor choice: Further evidence. Auditing Research, (5), 84–90 (in Chinese). Gong, Q. H., Liu, H. L., & Shen, H. H. (2010). The development of regional factor market, state holding and cost stickiness. China Accounting Review, (4), 431–445 (in Chinese). Gul, F. A., & Tsui, J. (1998). A test of the free cash ﬂow and debt monitoring hypotheses: Evi- dence from audit pricing. Journal of Accounting and Economics, 24, 219–237. Guo, Z. R. (2011). International big four accounting ﬁrms and high auditing quality: Evidence from China’s securities market. Auditing Research, (1), 98–107 (in Chinese). Hong, J. M., Xu, Y. D., & Li, Y. R. (2011). Information disclosure quality, controlling share- holder’s embezzlement and auditor selection. Auditing Research, (2), 107–112 (in Chinese). Jensen, M. C. (1986). Agency costs of free cash ﬂow, corporate ﬁnance, and takeovers. The American Economic Review, 76, 323–329. Jensen, M. C., & Meckling, W. H. (1976). Theory of the ﬁrm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3, 305–360. Jian, M., & Wong, T. J. (2010). Propping through related party transactions. Review of Account- ing Studies, 15(1), 70–105. Jiang, F. X., Huang, L., & Zhang, M. (2009). Product market competition, corporate governance and agency cost. The Journal of World Economy, (10), 46–59 (in Chinese). Jiang, R., Liu, X., & Liu, B. (2007). An empirical study on the governance effect of the exterior audit from Chinese listed companies: A view from CEO turnover. Journal of Finance and Economics, (11), 92–103 (in Chinese). Lawrence, A., Minutti-Meza, M., & Zhang, P. (2011). Can Big 4 versus non-Big 4 differences in audit-quality proxies be attributed to client characteristics? The Accounting Review, 86(1), 259–288. Lennox, C. S. (1999). Audit quality and auditor size: An evaluation of reputation and deep pock- ets hypotheses. Journal of Business Finance & Accounting, 26, 779–804. Li, S. X. (2007). Ownership, agency cost and agency efﬁciency. Economic Research Journal, (1), 102–113 (in Chinese). Li, W. A., Wang, X. H., & Wang, W. (2004). Earnings management and audit opinions: An empirical study based on non-operating earnings. Journal of Finance and Economics, (11), 126–135 (in Chinese). Liang, S. K. (2013). Ownership concentration and the viscosity of cost. Economic Research Jour- nal, (1), 144–155 (in Chinese). China Journal of Accounting Studies 321 Liu, F., & Xu, F. (2002). Risk based auditing, litigation risk and auditing quality. Accounting Research, (2), 21–27 (in Chinese). Liu, F., & Zhou, F. Y. (2007). Does size really matter? — A perspective of accounting conserva- tism test. Accounting Research, (3), 79–87 (in Chinese). Liu, J. H. (2009). Ownership, earning management and audit opinions. Auditing Research, (2), 32–39 (in Chinese). Liu, Q., & Yan, Z. Y. (2006). Accounting estimate and quality of independent auditing. Account- ing Research, (9), 82–88 (in Chinese). Ministry of Finance of the People’s Republic of China. (1993). Temporary Regulations on the Examination and Approval Procedures of Limited Liability Accounting Firms (in Chinese). Ministry of Finance of the People’s Republic of China. (1998). The Notice about Decoupling Accounting Firms which have Securities or Futures Related Business from their Linked Units (in Chinese). Moore, G., & Scott, W. (1989). Auditors’ legal liability, collusion with management. Contempo- rary Accounting Research, 5, 754–774. O’Keefe, T., & Westort, P. (1992). Conformance to GAAS reporting standards in municipal audits and the economics of auditing: The effects of audit ﬁrm size, CPA examination perfor- mance and competition. Research in Accounting Regulation, 6,39–77. Petersen, M. A. (2009). Estimating Standard errors in ﬁnance panel data sets: Comparing approaches. Review of Financial Studies, 22(1), 435–480. Qi, J. N., Chen, H. L., & Zhang, Y. (2004). Auditor size, brand, price and audit quality: Research on charging and quality of international big four Chinese audit marketing. Auditing Research, (3), 59–65 (in Chinese). Shleifer, A., & Vishny, R. (1986). Large shareholders and corporate control. Journal of Political Economy, 94, 461–488. Stulz, R. M. (1990). Managerial discretion and optimal ﬁnancing policies. Journal of Financial Economics, 26(1), 3–27. Sun, Z., & Liu, H. (2004). The expense ‘stickiness’ behavior of Chinese listed companies. Eco- nomic Research Journal, (12), 26–34 (in Chinese). Thornton, D. B., & Moore, G. (1993). Auditor choice and audit fee determinants. Journal of Business Finance & Accounting, 20, 333–349. Wan, S. Y., & Wang, H. J. (2011). Managerial self-interest, board governance and cost stickiness. Economy and Management, (3), 26–32 (in Chinese). Wang, J. C. (2011). Empirical study on the audit independence of top management turnover in the listed companies. Journal of Shanxi Finance and Economics University, (9), 116–124 (in Chinese). Wang, Q., Wong, T. J., & Xia, L. (2008). State ownership, the institutional environment, and auditor choice: Evidence from China. Journal of Accounting and Economics, 46(1), 112–134. Wang, X., & Zhang, W. G. (2005). Financial restatement and independent audit quality. Auditing Research, (3), 56–61 (in Chinese). Wang, Y. M., & Wang, P. (2006). Research on the difference in market reactions to auditor’s ﬁnancial statements between big-fours and non-big-fours. Auditing Research, (5), 49–56 (in Chinese). Watts, R. L., & Zimmerman, J. L. (1981). The markets for independence and independent audi- tors. Working Paper. Watts, R. L., & Zimmerman, J. L. (1983). Agency problems, auditing, and the theory of the ﬁrm: Some evidence. Journal of Law and Economics, 26, 613–633. Watts, R. L., & Zimmerman, J. L. (1986). Positive accounting theory. Englewood Cliffs, NJ: Prentice-Hall. Weiss, D. (2010). Cost behavior and analysts’ earnings forecasts. The Accounting Review, 85, 1441–1471. White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica, 48, 817–838. Williamson, O. E. (1963). Managerial discretion and business behavior. The American Economic Review, 53, 1032–1057. Wu, X. (2006). Accounting ﬁrm combination and quality control: The case of zhongtianqin. Accounting Research, (10), 79–85 (in Chinese). 322 Liang et al. Yang, D. M., Lin, B., & Wang, Y. C. (2009). Internal control, audit quality and large shareholder tunneling. Auditing Research, (5), 74–81 (in Chinese). Yu, P. (2007). Firm characteristics, the big-four and audit opinions. Auditing Research, (2), 53–60 (in Chinese). Yuan, H. Q., & Li, H. J. (2003). The organization form, scale and audit quality of accounting ﬁrms. Auditing Research, (1), 32–37 (in Chinese). Yue, H. (2006). Large shareholder tunneling and the monitoring role of auditor. China Accounting Review, (6), 59–68 (in Chinese). Zang, A. Y. (2012). Evidence on the trade-off between real activities manipulation and accrual-based earnings management. The Accounting Review, 87, 675–703. Zeng, Y., & Ye, K. T. (2005). Ownership structure, agency cost and external auditing demand. Accounting Research, (10), 63–70 (in Chinese).

Journal

China Journal of Accounting Studies – Taylor & Francis

Published: Oct 2, 2014

Keywords: Big Four International; cost stickiness; shareholding ratio; marketisation level; propensity score matching

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

External auditor types and the cost stickiness of listed companies

External auditor types and the cost stickiness of listed companies

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

External auditor types and the cost stickiness of listed companies

External auditor types and the cost stickiness of listed companies

References (70)

Abstract

Journal

Recommended Articles

There are no references for this article.

Our policy towards the use of cookies