Determinants and consequences of voluntary disclosure in an emerging market: Evidence from China

Transcription

1 Available online at Journal of International Accounting, Auditing and Taxation 17 (2008) Determinants and consequences of voluntary disclosure in an emerging market: Evidence from China Kun Wang a,, Sewon O a, M. Cathy Claiborne b a Department of Accounting & Finance, Texas Southern University, Houston, TX 77004, United States b University of Colorado at Colorado Springs, Colorado Springs, CO 80918, United States Abstract In this paper we examine empirically the determinants of voluntary disclosure in the annual reports of Chinese listed firms that issue both domestic and foreign shares and determine if the cost of debt capital is related to the extent of voluntary disclosure. We find the level of voluntary disclosure is positively related to the proportion of state ownership, foreign ownership, firm performance measured by return on equity, and reputation of the engaged auditor. There is no evidence, however, that companies benefit from extensive voluntary disclosure by having a lower cost of debt capital Elsevier Inc. All rights reserved. Keywords: Voluntary disclosure; Emerging market; Stock ownership; Auditor type 1. Introduction Voluntary disclosure, i.e., information in excess of mandatory disclosure, in the context of globalization of the world s financial markets has received considerable attention in the accounting literature in recent years. However, much of the research to date has focused on developed countries (Cooke, 1989; Meek & Gray, 1989; Meek, Roberts, & Gray, 1995). Little attention has been devoted to the voluntary disclosure of companies in China, a unique economy that has gained increased importance in the global capital market. China is worthy of study for several reasons. First, China is one of the fastest growing economies in the world and one of the world s largest exporters/importers. In the past decade, China s stock market capitalization and daily turnover has grown from nothing to 9 trillion RMB (US$ 1.12 trillion) and 84 billion RMB (US$ 10.5 billion), respectively. In 2006 the Chinese stock market was the third largest market in Asia after Tokyo and Hong Kong. Second, with the development of its securities market, the Chinese government has undertaken steps to improve the Chinese investment climate to make it more appealing to domestic and foreign capital funds. Finally, in the current liberalized economic climate, Chinese firms are also taking steps to attract overseas investors. To this end, listing on overseas stock exchanges and attracting overseas investors to invest directly in their shares in the Chinese market are steps taken by local firms in China. In an effort to achieve competitive advantages, Chinese firms have been adopting new disclosure strategies, including increased voluntary disclosure in annual reports. These developments will have a profound impact not only on the Chinese financial markets, but on the U.S. and global markets as well. Our study is, therefore, expected to shed light on Corresponding author. addresses: wangk@tsu.edu (K. Wang), sewono@tsu.edu (S. O), cclaibor@uccs.edu (M.C. Claiborne) /$ see front matter 2008 Elsevier Inc. All rights reserved. doi: /j.intaccaudtax

2 K. Wang et al. / Journal of International Accounting, Auditing and Taxation 17 (2008) corporate information disclosure in a growing capital market. Specifically, this paper seeks to examine the determinants of voluntary disclosure for Chinese public firms and its association with the cost of debt capital. The disclosure index used in this study is based on those used in earlier studies (Gray, Meek, & Roberts, 1995) and has been tailored to reflect the features of voluntary disclosure in the Chinese context. The sample includes the annual reports of all Chinese listed companies that have issued both domestic and foreign shares as of December 31, We find firms tend to expand voluntary disclosure when they have a higher proportion of ownership by stateowned representatives and foreign investors. These results reflect a specific feature of the information disclosure in the Chinese setting. We also find a significant positive relation between voluntary disclosure and a company s financial performance measured by return on equity. Further, employing a Big 4 auditor substantially increases the amount of corporate information disclosed to the public. Additional analysis, however, offers no evidence that companies can benefit from extensive voluntary disclosure by having a lower cost of debt capital. This finding is consistent with the underdeveloped nature of the Chinese debt market that is making the transition from a planned economy to a market economy. Our study adds to the growing literature on international disclosure practices and their determinants. The study also benefits investors because knowing disclosure characteristics will help investors find desired information about Chinese companies. Investors have to rely more on analysts and information other than a firm s annual report for firms with less state and foreign ownership, firms with poor performance, or firms audited by a non-big 4 auditor. Finally, the study provides insights for accounting standard setters, especially the Chinese Securities Regulatory Commission, in answering questions such as What types of firms should be subjected to disclosure regulations? and Is there a need for certain forms of regulation when firms disclose voluntarily? The remainder of the paper is structured as follows. The next section offers background information about the corporate disclosure regime in China, and develops hypotheses based on the information asymmetry theory. Section 3 explains the research methodology. Empirical results are provided in Section 4. The final section presents conclusions and limitations. 2. Background and hypotheses 2.1. Background Until the late 1970s, China operated a planned economic system under which an enterprise was either state-owned or collectively owned. Both types of enterprise were run directly by the government, with little room for market mechanisms. Government agencies and company management were the only users of financial reports. Economic reform and adoption of an open door policy, however, have resulted in dramatic changes to the disclosure environment. The most important change is the emergence of new forms of business including private companies, Sino-foreign joint-ventures, foreign companies, and share capital-based companies. These new types of business have brought new users of corporate information, such as investors and financial analysts. Recently, domestic capital markets have been re-energized by the so-called G-share reforms that permit more government-owned shares to trade and by domestic IPOs. The second major change has involved the establishment of the Shanghai Stock Exchange and Shenzhen Stock Exchange in 1990 and 1991, respectively. The two markets have expanded steadily and provide direct incentives and pressures for market-oriented financial disclosure. 1 At the end of February 2007, 1446 companies listed their shares in the two stock markets. Currently, two types of shares are listed on the domestic exchanges: A-share and B-share. A-share listings are offered only to domestic investors and are transacted in Chinese currency (RMB). B-share listings are offered, primarily, to foreign investors and transacted in US dollars (Shanghai) or HK dollars (Shenzhen). 2 Approximately 20% of A-share firms are also authorized to issue B-shares. Since its inception in 1991, the B-share market has attracted a considerable number of foreign investors. By the end of November 2006, there were a total of 110 B-share issuers with total market capitalization of US$ billion. 1 The State Planning Commission determines an annual quota for listing on the domestic and international exchanges. 2 Until 2001 B-shares were only offered to foreign investors.

3 16 K. Wang et al. / Journal of International Accounting, Auditing and Taxation 17 (2008) Financial information disclosure by Chinese companies has been the subject of prior research. Liu and Zhang (1996) conducted descriptive research of the mandatory disclosure level of Chinese enterprises based on the disclosure requirements of the government. They reported a high level of compliance with the mandatory disclosure requirements, with government monitoring playing an important role in achieving this outcome. Many voluntary disclosures also were found, especially in earnings forecasts and supervisory board s reports. Ferguson, Lam, and Lee (2002) examined, empirically, the impact of international capital market pressure on the voluntary disclosure in the annual reports of formerly wholly state-owned enterprises (SOEs) listed on the Stock Exchange of Hong Kong (SEHK). Consistent with a cost benefit framework, they found that SOEs disclosed significantly more strategic and more financial information than other SEHK firms. More recently, Xiao, He, and Chow (2004) analyzed the factors behind Chinese listed companies voluntary adoption of internet-based financial reporting, as well as the extent of their disclosure. Factors identified as being relevant to voluntary disclosure choices include auditor type, foreign listing, stock ownership, and government regulation. Our study extends the prior research in two aspects. First, we are particularly interested in the extent of voluntary disclosures in the annual reports of Chinese listed companies that issue both A- and B-shares. Aside from attempting to attract foreign capital by directly listing in foreign exchanges, Chinese firms issue B-shares to foreign investors as an alternative to raise additional capital. Similar to companies with foreign listing status, B-share issuers have taken a variety of actions to protect investors interest, increase the transparency of the B-share market, and improve disclosure of information to achieve a competitive advantage. For example, B-share companies are required to prepare their annual reports in accordance with both the Chinese and the International Financial Reporting Standards. As suggested by agency theory, firms with dual listing status (foreign and domestic), such as B-share companies, are extremely motivated to disclose more voluntary information as compared to firms that only issue domestic shares, and as such, provide us with a unique setting to explore the characteristics affecting corporate information disclosure in emerging economies. Second, in addition to identifying determinants of voluntary disclosure, we explore the association of voluntary disclosure and the companies cost of debt capital. To date, empirical evidence from U.S. markets is consistent with the view that voluntary disclosure reduces information asymmetry and facilitates a firm s access to lower-cost external financing (Botosan, 1997; Healy, Hutton, & Palepu, 1999). However, evidence on this research question is scarce in an international setting such as China where an investor protection environment is weak and financial structure is still heavily bank-orientated. Additionally, unlike the commercial banking system in mature economies, banks in China traditionally met government policy goals by financing the operations of previous state-owned enterprises (SOEs), regardless of their profitability or risk. As a result, they functioned below potential by incurring bad debt due to the lagging performance of SOEs. Following the market economy reform, Chinese authorities have attempted to ease this constraint by implementing policies that increase the role of market forces and encourage banks to engage in commercial banking, and by privatizing previously stated owned banks. In this regard, research is needed to determine any disclosure-cost of capital relationships in a developing economy. Also an assessment of additional incentives for companies to engage in information disclosure is likely to be insightful Hypotheses In a discussion of the management of both quantitative and qualitative corporate financial disclosures, Gibbins, Richardson, and Waterhouse (1990) point out that companies develop disclosure strategies in response to both internal and external conditions. Therefore, the firm s disclosure decisions are likely to be motivated by a variety of factors, e.g., agency costs (Leftwich, Watts, & Zimmerman, 1981); litigation costs (Skinner, 1994); information asymmetries (Hughes, 1989); and disclosure related costs (Ali, Ronen, & Li, 1994). In this study we draw on the framework of information asymmetries/signaling and their association with information disclosure to develop empirical predictions. Theory (Grossman, 1981; Spence, 1973) indicates that voluntary disclosure can be used to alleviate information asymmetry problems, including moral hazard and adverse selection. A rational strategy to avoid deep discounting of share prices is to disclose additional information to investors to signal firm value (Watts & Zimmerman, 1986). Taken together with factors specific to the Chinese context, we generate our hypotheses from the following aspects.

4 K. Wang et al. / Journal of International Accounting, Auditing and Taxation 17 (2008) State ownership The Chinese stock market emerged with the implementation of privatization of previous state-owned enterprises, as a means to improve productivity and corporate governance and to raise much needed capital in local and international markets. Prior to economic reform, SOEs were characterized by a lack of autonomy and marginal effect of managers incentives on profitability. To improve efficiency, the Chinese government took a series of actions to prompt privatization of SOEs and as a result, adopted the current corporate system under which the state is entitled to a dividend on its share in SOEs assets. Shares owned by the government represent 50% of the total untradable shares. Increased management autonomy accompanied by the need to raise capital externally, however, led to the need for improved governance structures. The listing of shares on both domestic and international exchanges is viewed as an essential part of this reform. Compared to other types of listed firms, Ferguson et al. (2002) argue that SOEs are likely to present significantly greater adverse selection and moral hazard problems. For example, unlike the westernized corporate governance systems in other business forms, many SOEs still operate in a vacuum with respect to corporate governance and management control (World Bank, 1995). Therefore, SOEs face significantly greater incentives to voluntarily disclose additional information to ease investor concerns regarding management quality, the potential for asset stripping or misappropriation, and the role of the government as a major shareholder. In the meantime, additional disclosure by SOEs is also likely to be less costly since most of them operate in industries deemed by the Chinese government to be of strategic importance and are hence shielded from international competition. Consistent with their hypothesis, Ferguson et al. (2002) find SOEs that are dual-listed in the Stock Exchange of Hong Kong disclose significantly more strategic and financial information than domestically listed firms. In rebuttal, another research by Xiao et al. (2004) report that companies with higher state ownership reduce, rather than increase their website information disclosure to the public. They interpret this result as a reflection of the current lack of emphasis on efficiency and profitability by state shareholders or their direct access to corporate insider information. Taking the prior evidence together, it appears that the sign of the association between the extent of voluntary disclosure and state ownership is undeterminable. After carefully analyzing our sample, we find the composition and characteristics of the companies are more like dual-listed SOEs in Ferguson et al. (2002) than the 300 largest domestically listed firms investigated by Xiao et al. (2004). Specifically, 71% of the companies are former wholly owned SOEs and the average state ownership is 27%. Given the government s keen interest in obtaining foreign capital, it is reasonable that voluntary disclosure of additional information is encouraged for B-share firms. Consequently, we leave the state ownership disclosure relationship as an open hypothesis to be tested, but expect a positive association to be dominant. H1. The extent of voluntary disclosure for B-share companies is related to the proportion of state ownership Foreign ownership Agency theory (Fama & Jensen, 1983) suggests that as the number of shareholders increases and ownership becomes more dispersed, monitoring costs, and hence demands for additional information, increase. Recent studies show that there is a complementary (positive) association between foreign listing status and the extent of voluntary disclosure. For instance, Meek and Gray (1989) found that continental European multinational corporations listed on London Stock Exchange voluntarily disclosed information in excess of what was required under London Stock Exchange rules. Likewise, Hossain, Tan, and Adams (1994) documented that Malaysian multinationals listed on London Stock Exchange voluntarily disclosed more information in their annual reports than companies listed only on local stock exchanges. In the Chinese setting, foreign shareholders of listed companies are likely to face higher levels of information asymmetry given their language barrier and lack of access to corporate information. It also implies Chinese firms have to supply more transparent disclosure that is particularly important for foreign investors in order to raise and retain foreign funds. In this context, Ferguson et al. (2002) report that Chinese firms quoted on several stock exchanges make more information disclosure. Xiao et al. (2004) find higher foreign ownership not only encourages information disclosure, but also motivate firms to create English web pages to facilitate dissemination of financial information. B-share firms are also required to follow International Financial and Reporting Standards by the Chinese Securities Regulation Commission (CSRC). This additional requirement is aimed to improve the overall disclosure and transparency as well as investor relations. In this regard, we predict a positive association between foreign ownership and the extent of corporate voluntary disclosure and develop our second hypothesis:

5 18 K. Wang et al. / Journal of International Accounting, Auditing and Taxation 17 (2008) H2. The extent of voluntary disclosure for B-share companies increases with higher proportion of foreign ownership Firm performance Firm performance is time specific and represents information that may be of interest to accounting information users (Wallace, Naser, & Mora, 1994). Singhvi and Desai (1971) claimed that in the face of adverse selection, higher profitability might induce management to supply more information to illustrate its ability to maximize the shareholders value and to increase its managerial compensation. Similarly, management of a profitable firm may feel proud of its achievement and wish to disclose more information to the public to promote a positive impression of its performance. In contrast, Lang and Lundholm (1993) contended that disclosure is influenced by a company s relative performance and hence, the direction of relationship between the performance and disclosure is rather unclear. For example, firms with negative information (particularly earnings information) might wish to convey more information to enhance creditability or to reduce the likelihood of legal liability. To date, the empirical evidence on the relation between firm performance and disclosure is mixed (Camfferman & Cooke, 2002; Lang and Lundholm, 1993; Wallace et al., 1994). In China, due to the short history of the stock market and lack of financial intermediaries, many investors, especial small ones, rely heavily on financial information disclosed in a company s financial reports to make investment decisions. As more and more new issues become available to the public, earnings performance inevitably becomes a strong factor in selecting high quality firms from the lemon companies. In response to investors fixation on earnings as well as external incentive mechanism, management of profitable companies are motivated to provide broad and deep analyses of financial conditions, expecting to reduce agency cost and distinguish themselves. In contrast, management experiencing lower profitability or loss normally feel threatened and wish to obscure poor results by disclosing less information. A study of Fan (2006) lends support to these arguments by indicating Chinese companies listed in the Shenzhen Stock Exchange disclose significantly more information when they achieve higher return on equity for the current period. Taken as a whole, we include return on equity as a performance measure and our third hypothesis states: H3. The extent of voluntary disclosure for B-share companies increases with higher return on equity Auditor type The signaling literature suggests that the choice of an external auditor can serve as a signal of firm value. Generally, entrepreneurs are likely to choose a large audit firm (e.g., Big 4) since such an action signals to investors their acceptance of the auditor s demands for higher quality disclosure as well as the quality of a firm s earnings performance (Datar, Feltham, & Hughes, 1991). This expectation is also consistent with agency theory that holds large auditors have a stronger incentive to maintain independence and to impose more stringent and extensive disclosure standards because they have more to lose from damage of their reputations. Unlike the western countries (e.g., U.S., U.K., Australia), the Chinese audit market has been proven to be competitive due to lower market concentration of Big 4 firms and active participation of second tier and small auditors (Li, Song, & Wong, 2005). In addition, the critical role of auditors in fostering transparent company reports and maintaining market confidence are widely recognized by both the regulators and other market participants. As a result, the demand for integrity and high quality audits creates a potential opportunity for Big 4 firms to serve as a signaling device to attract investment funds. The findings of Xiao et al. (2004) support this proposition with a positive relationship between firms employing former Big 5 firms and their scopes of voluntary disclosure. Moreover, CSRC recommends a Big 4 audit firm for B-shares firms in order to increase the credibility of their annual reports. Given the general inference and unique features of the B-share market, we expect a positive association between a large audit firm and the extent of voluntary disclosure. This association is tested by our fourth hypothesis: H4. The extent of voluntary disclosure for B-share companies increases if they are audited by a Big 4 accounting firm. Regarding the economic consequences of voluntary disclosure, U.S. empirical evidence is consistent with the view that voluntary public disclosure reduces information asymmetry and facilitates a firms access to lower-cost external financing. For example, a negative relationship exists between a firm s level of voluntary disclosure and its cost of debt capital (Sengupta, 1998) and equity capital (Botosan, 1997). Using a group of international companies, Francis, Khurana, and Pereira (2005) find firms benefit from expanded disclosure by having a lower cost of both debt and equity

6 K. Wang et al. / Journal of International Accounting, Auditing and Taxation 17 (2008) Table 1 Industry representation and auditor engagement of the sample firms Industry description Number of firms % of the sample Panel A: Industry representation of the sample firms Agriculture Chemicals and allied products Conglomerates Construction Electronics Food and beverages Industrial and commercial machinery IT Mining and metal productions Paper and printing Pharmaceuticals Real estate Social services Textiles and apparel Transportation Utilities Wholesales and retails Total sample IPOs Audit firms # of domestic appointments % # of foreign appointments % Panel B: Auditor engagement of the sample firms Ernest & Young Deloitte & Touche KPMG Peat Marwick Price Waterhouse & Coopers BDO Horwath International Morison Grant Thornton Other local CPA firms Total sample IPOs 108 a a 100 a The numbers are different from the total sample due to missing auditor data. capital after controlling for cross-country institutional differences in legal and financial systems. Since China is still in the transitional stage from a bank-oriented to market-orientated financial system, it remains an empirical question whether the predictions derived from the extant theory hold for Chinese public firms, especially for the cost of debt capital and voluntary disclosure linkage. The research discussed above leads to our final hypothesis: H5. The cost of debt capital decreases as B-share companies increase the level of voluntary disclosure. 3. Research methodology 3.1. Sample selection Our study includes all of the 110 Chinese listed companies that issued both A- and B-shares at the end of One company was deleted due to a missing annual report. The voluntary disclosure information was hand collected from each company s annual report published by the Shanghai and Shenzhen Stock Exchanges. The companies industry representations are reported in panel A of Table 1. Industrial and commercial machinery accounts for 27.52%, textile and apparel 9.17%, and electronics and transportation 8.26% each of the population. The remaining 46.75% is from 14 other sectors.

7 20 K. Wang et al. / Journal of International Accounting, Auditing and Taxation 17 (2008) Since we are interested in the auditors impact on the disclosure practice, Panel B of Table 1 offers market share information for Big 4 accounting firms and other major international auditors. 3 For the domestic (foreign) financial reports, total market share of Big 4 auditors is 19.45% (34.58%). This is consistent with the prior research, suggesting the Chinese audit market is competitive rather than oligopolistic (Li et al., 2005). In comparison, the markets for publicly listed firms of the US, UK, Canada, and Hong Kong were dominated by the Big 8/5/4. In these countries, the big firms audited more than 80% of listed firms and captured more than 97% of sales revenue. In terms of market profiles for individual audit firms, Price Waterhouse & Coopers engages more clients in both the domestic and foreign markets than the other Big 4 firms (10.19% and 19.63%, respectively), followed by Deloitte and Touche and KPMG. Noticeably, two second tier auditors, namely BDO and Horwath, have larger combined market shares than Big 4 firms. The underlying reason is that in an effort to increase market share, international auditors especially the second tier players have accelerated their speed to merge or form partnerships with Chinese local CPA firms. For example, Horwath International has formed partnerships with seven local firms since 2001, including Shanghai Shulun Pan CPAs, one of the largest local CPAs in China at that time. The potential effects of these non-big 4 auditors on our hypotheses will be addressed in the next section Disclosure index Cooke and Wallace (1989) consider financial disclosure to be an abstract concept that cannot be measured directly. Nevertheless, they concede that a suitable proxy such as an index of disclosure can be used to gain insight into the level of information disclosed by companies. Accordingly, an index of disclosure was employed in this study to gauge the extent of voluntary disclosure by companies. The 2005 annual reports were obtained from the companies selected and formed the latest source of information available at the time the study was conducted. A number of steps were taken to develop the index of discretionary items. These steps were: (1) based on the disclosure index used in earlier studies (Cooke, 1989; Ferguson et al., 2002; Gray et al., 1995) a preliminary list of 93 items was generated; (2) the list was then examined by referring to the Chinese accounting standards in order to develop an index of discretionary items. Fourteen items were dropped from the list because they were mandatory disclosure requirements. After the index was revised, a final list of 79 discretionary items was compiled. Approximately 70% (56) of the disclosure index items is similar to those selected by Gray et al. (1995). After establishing the disclosure index, a scoring sheet was developed to assess the extent of voluntary disclosure. If a company disclosed an item of information included in the index, it received a score of 1, and 0 if it is not disclosed (see Cooke, 1989, p. 182). Each firm s voluntary disclosure score was further partitioned into (1) strategic, (2) non-financial, and (3) financial information based on the model developed by Meek et al. (1995). The disclosure index partitioned by information type is presented in Appendix A. The disclosure index is unweighted and assumes that each item of disclosure is equally important (Gray et al., 1995). Cooke (1989, p. 182) suggests that unweighted indices are an appropriate research instrument in disclosure studies when the focus of the research is directed at all users of corporate annual reports rather than the information needs of any specific user group. Furthermore, the weighted indices are subject to widespread criticism in the literature. For example, Chow and Wong-Boren (1987) argue that a great deal of subjectivity exists in the assignment of weights because it reflects the perceptions of information needs rather than actual information needs of the users of financial reports. Since this study is not directed at a single user group, an unweighted index is considered appropriate. The overall index derived by this procedure is expressed in two ways. One is the raw count of the disclosure items and the other is a proportion of the actual score allocated to a company to the maximum possible score applicable to that company. Consequently, companies were not penalized for non-disclosure of an item if it was deemed to be irrelevant to their business activities. As with prior research (e.g., Cooke, 1989), the entire annual report was read to assess the relevance of a particular item of information to the firm. The tests of H1 H4 are based on the following model: DRAW (DSCORE) = MKTVAL + LEV + STATESH + FOREIGN + AUDITOR + ROE + e (1) 3 China announced plans to develop its own international accounting firms in the next 5 10 years.

8 K. Wang et al. / Journal of International Accounting, Auditing and Taxation 17 (2008) Market value (MKTVAL) is a control variable for firm size in the model. A number of disclosure studies find that firm size is an important factor in explaining variability in the extent of corporate voluntary disclosure. Academicians identify a positive relationship between firm size and voluntary disclosure in the setting of U.S. (Firth, 1979), Swedish (Cooke, 1989), Malaysia (Hossain et al., 1994) and Japanese firms (Cooke, 1991), as well as for firms listed on multiple exchanges (Meek et al., 1995). Another control variable in the model is leverage (LEV). Arguably, agency costs, and hence disclosure, will be higher for firms with proportionately more debt because of the increased potential for wealth transfers from debt holders to shareholders and managers (Jensen & Meckling, 1976). However, research findings on this relation are mixed. Sengupta (1998) found that U.S. firms with high analyst disclosure quality ratings enjoyed lower cost of issuing debt. Similarly, Bradbury (1992) found a significant, positive relationship between leverage and disclosure for New Zealand firms. Chow and Wong-Boren (1987), however, found no relationship between leverage and disclosure in their sample of Mexican firms, while Meek et al. (1995) report a significant, negative relationship between leverage and voluntary disclosure for U.S., U.K., and continental European multinationals. To examine the effects on different types of disclosure, Eq. (1) was replicated with each firm s voluntary disclosure score partitioned into additional strategic, non-financial, and financial information Cost of debt analysis Since researchers are generally interested in the economic consequences of voluntary disclosure, our study performs additional tests to determine whether extensive voluntary disclosure influences firms cost of debt capital in external financing (H5). To implement this test, we use a model developed by Francis et al. (2005) that is expressed as follows: COD = DSCORE (SSCORE, NFSCORE, FSCORE) + LEV + TA + ROA (2) Table 2 provides a summary of all of the variables used in the hypothesis tests. The independent variables are measured using data obtained from the company s annual reports. 4. Results and analyses 4.1. Empirical results of the voluntary disclosure tests (H1 H4) Descriptive statistics Panel A of Table 3 presents descriptive statistics for the variables in the analysis. The median values of DRAW and DSCORE are 13 and 0.18, respectively. Consistent with prior findings, there is considerable variation in DRAW, as represented by its minimum, maximum, and standard deviation. Compared to the median value 0.21 reported by Ferguson et al. (2002) for Chinese firms dual-listed in Hong Kong, the overall disclosure of our sample is slightly lower. Regarding the partition of the overall disclosure, the median values for strategic and financial information are 0.32 and 0.14, respectively. These values are comparable to those reported by Ferguson et al. (2002). Nevertheless, we find firms in our sample are substantially inferior when it comes to communication of non-financial information to the market, such as employment policy and environment protection plan. Only 12 firms (11% of the total sample) disseminate non-financial items and the mean score of 0.01 is very low. This result may be due to market regulations and/or public scrutiny since there is substantially higher pressure perceived by dual-listing Chinese firms to disclose non-financial information in the Hong Kong market. As non-financial information is one of the critical factors used by investors to assess a company s sustainability in the long run, the low non-financial score also implies a potential area in which Chinese regulatory bodies need to enact further enforcements to promote a transparent market. The descriptive statistics in Table 3 also reveal some interesting patterns. The mean (median) assets of the sample firms is 4188 (2417) million Chinese RMB (US$ 524 and US$ 303, respectively), and there is considerable dispersion within the sample. While the state ownership (26.84%) is close to that reported in Xiao et al. (2004), the foreign ownership (33.71%) is much larger than the 8% in their study because our sample contains exclusively B-share companies. After collapsing domestic and international auditor into one variable BIG4, we find 36% of the firms are audited by Big 4 auditors. Finally, the median of leverage and return on equity are 0.56 and 4.45%, respectively, and

9 22 K. Wang et al. / Journal of International Accounting, Auditing and Taxation 17 (2008) Table 2 Explanations of dependent and independent variable Panel A: The voluntary disclosure regression Dependent variables DRAW DSCORE SSCORE NFSCORE FSCORE Test variables State ownership (STATESH) Foreign ownership (FOREIGN) Return on equity (ROE) Type of auditor (AUDITOR) Control variables Firm size (MKTVAL) Leverage (LEV) Panel B: The cost of debt capital regression Dependent variables Cost of debt (COD) Test variables DSCORE SSCORE NFSCORE FSCORE Control variables Firm size (TA) Leverage (LEV) Return on assets (ROA) Number of actual disclosure items Proportion of the actual score allocated to a company to the maximum possible score applied Proportion of the actual strategic disclosure score allocated to a company to the maximum possible score applied Proportion of the actual non-financial disclosure score allocated to a company to the maximum possible score applied Proportion of the actual financial disclosure score allocated to a company to the maximum possible score applied Percentage of state-owned shares of a company Percentage of foreign investors shares of a company Net income divided by the shareholders equity A binary variable which took the value of 1 for Big 4 audit firms and 0 for non-big 4 audit firms Log of the companies market value Ratio of total debt to owners equity Interest expenses divided by total debts Same as Panel A Same as Panel A Same as Panel A Same as Panel A Log of the companies total assets Ratio of total debt to owners equity Net income divided by total assets their means are not fully reprehensive of the data because of the extreme values. To remove the potential effects of these outliers, we winsorized the top and bottom two percent observations in the regression analysis. Panel B of Table 3 reveals a number of significant correlations between pairs of key variables. These suggest the potential for at least some of the hypotheses to be supported. MKTVAL has significant correlations with DSCORE. ROE is significantly correlated with DSCORE. There is also significant correlation between BIG4 and DSCORE. However, we find that the correlations between STATESH and DSCORE, and between INDFOR and DSCORE are not significant. Panel B of Table 3 also shows that among the independent variables, correlations are relatively low. To further assess the potential for multicollinearity, we preformed regressions of all explanatory variables on DSCORE (not reported), and obtained variance inflation factors (VIF) below 1.3 and tolerance levels above 0.82 for all independent variables. The results suggest that multicollinearity between the independent variables is unlikely to pose a serious problem in the interpretation of the results of the multivariate analysis Multivariate analysis The regression results of the voluntary disclosure model are displayed in Table 4. We use both the raw disclosure counts (DRAW) and the disclosure ratios (DSCORE) as the dependent variables and obtained similar results. 4 In the 4 An ordinary Least Square (OLS) model was used for the DSCORE regression. Since the dependent variable is a proportion which is bounded by 0 and 1, an arcsine-toot transformation of the dependent variable is applied to get unbiased estimates (Hopkins, 2000). The White and Breusch-Pagan tests detect no heteroscedasticity in any of these OLS models.

10 K. Wang et al. / Journal of International Accounting, Auditing and Taxation 17 (2008) Table 3 Summary statistics for the voluntary disclosure model Variables Mean Median Standard deviation Minimum Maximum Panel A: Descriptive statistics DRAW DSCORE SSCORE NFSCORE FSCORE MKTVAL TA LEV STATESH FOREIGN ROE BIG Variables DSCORE SSCORE NFSCORE FSCORE MKTVAL LEV STATESH FOREIGN ROE BIG4 Panel B: Correlation matrix DSCORE 1 SSCORE 0.41* 1 NFSCORE 0.31* FSCORE 0.6* * 1 MKTVAL 0.29* ** 0.28* 1 LEV 0.23** STATESH FOREIGN *** 0.25* 1 ROE 0.28** * 0.21** 0.19* BIG4 0.23* * 0.32* *** 1 *, **, *** indicate the correlation is significant at the 0.01, 0.05, and 0.10 level, respectively. Table 4 Regression results of the voluntary disclosure model Expected signs Overall disclosure Strategic information Financial information Goodness of fit Chi-square Significance Intercept? ** 3.88** 0.42 Independent variables a MKTVAL ** 4.54** 2.53 LEV STATESH ** 4.94** 0.01 FOREIGN ** 6.65* 1.61 ROE *** ** BIG *** * *, **, *** indicate significance at the 0.01, 0.05, and 0.10 level, respectively. a The first number is the coefficient estimate, and the second is the Chi-square of significance of the coefficient estimate.

11 24 K. Wang et al. / Journal of International Accounting, Auditing and Taxation 17 (2008) table we reported the Poisson regression for DRAW. These models were derived from crossing three dependent variables, the overall disclosure as well as disclosures from strategic and financial information. We omitted the non-financial information since there are very few firms disclosing non-financial items. The Chi-square statistics for goodness of fit tests support the significance of the models, implying that the independent variables explain a significant portion of the variance in the disclosure index. Among the control variables, MKTVAL is positively and significantly associated with all the disclosure measures, consistent with the size-disclosure relation documented in prior studies by Ferguson et al. (2002) for Chinese firms listed in Hong Kong, and by Xiao et al. (2004) for Chinese companies Internet financial disclosure. There are several reasons for this result, but most importantly, large firms are closely watched by investors and have the ability to absorb extra costs for broader disclosure. In contrast, LEV does not provide an explanation of the disclosure level variation in the Chinese context. The observation is not surprising as evidence from earlier studies is mixed, as discussed previously. Specifically, the outcome of this study is consistent with the findings of Chow and Wong-Boren (1987). Turning to the hypotheses, the coefficients for state-owned shares (STATESH) are positive and significant (p < 0.05) for the overall disclosure and strategic information but insignificant for the financial information. These findings suggest that Chinese B-share firms, with more ownership controlled by government, tend to disclose more information voluntarily to the market than firms with less or no government-owned shares. Ferguson et al. (2002) offered two reasons why Chinese cross-listed SOEs have incentives to disclose more information voluntarily. First, higher disclosure by these firms may reflect the effect of state-initiated disclosure policies. That is, the state is interested in developing good reputation through current SOEs so that future listings can raise more capital and are exposed to more investors. Second, management s disclosure decisions are influenced by their assessment of the associated costs and benefits as they attempt to reduce information asymmetry. While we cannot differentiate which of the two reasons applies to B-share firms based on our current data, we expect that both of them are playing important roles in promoting extensive corporate disclosure. This is because in the restructuring of SOEs, the government has put equal emphasis on attracting international capital and encouraging management s autonomy. As hypothesized in H2, all of the FOREIGN s coefficients are positive. The coefficients for overall disclosure and strategic information are significant (p < 0.05 and p < 0.01). This finding suggests that foreign shareholders play a positive role in monitoring management, as claimed by prior studies which found foreign ownership encourages the disclosure of a company s information through the Internet (Xiao et al., 2004). It also implies that financial statement users could rely more on the annual report to access corporate information if a company has higher percentage of foreign ownership. However, it is worth noting that while the two ownership variables (state and foreign) significantly affect the overall disclosure and strategic information disclosure, their impact on the financial information disclosure seem to be minimal. These results suggest that greater shareholdings by state and foreign owners do not induce companies disclosures beyond those mandated by the regulatory agencies. The other test variables also exhibit a pattern of selective impacts. The hypothesized effect of ROE is supported by the positive and significant coefficients on ROE in the overall disclosure and financial information models. These findings indicate that as the firm s earnings increase, managers have incentives to supply more information to the market in order to signal quality. On the other hand, investors can leverage more on voluntary disclosure to differentiate high quality stocks from the lemon companies. The impact of ROE is strongest for the financial disclosure and weak for the strategic information. This result implies that profitable firms are inclined to elaborate on financial analysis but hesitate to offer details about their operating environment and future development plans. We interpret this finding as a reflection that companies disclosure is triggered by investors fixation on short-term returns rather than long-term performance. Finally, H4 is also supported by BIG4 s positive and significant coefficients for the overall disclosure and financial information models. It appears that in the Chinese context, large independent auditors are playing a monitoring role that leads to increased voluntary disclosure as they do in advanced economies. Since the second tier auditors also have large market shares, we performed additional tests to explore their impacts of on the disclosure practice (not reported). First, we combine the two largest second tier auditors (Horwath and BDO) with Big 4s to form a new dummy variable but did not find significant influences on disclosure scores from this larger group of auditors. We further ran the regression analysis with a second tier auditor dummy in addition to BIG4. This time we still find no support for a second tier auditor effect while the BIG4 remains positively significant. Together, these results indicate it is the group of Big 4 auditors, not the second tier CPAs, that pushes more transparent disclosure from the companies. Despite the rigorous

12 K. Wang et al. / Journal of International Accounting, Auditing and Taxation 17 (2008) Table 5 Summary statistics for the cost of debt capital model Variables Mean Median Standard deviation Minimum Maximum Panel A: Descriptive statistics COD DSCORE SSCORE NFSCORE FSCORE TA LEV ROA Variables COD DSCORE SSCORE NFSCORE FSCORE TA LEV ROA Panel B: Correlation matrix COD 1 DSCORE SSCORE * 1 NFSCORE * FSCORE * * 1 SIZE * 0.17*** 0.26* 0.22** 1 TA ** * 1 ROA * *** 0.30* 0.40* -0.87* 1 *, **, *** indicate the correlation is significant at the 0.01, 0.05, and 0.10 level, respectively. competition in the Chinese setting, Big 4 firms appear to be able to maintain their reputations and assure quality. On the other hand, that the second tier auditors enjoy larger market share does not prove that their reputation and audit quality are improved proportionally. In fact, how to retain and increase uniform quality standards with more diversified partners appears to be a tough challenge for the second tier firms since it is quality that drives their long-run success Empirical results of the cost of debt analysis (H5) Table 5 provides descriptive statistics and correlation matrix of the interested variables in the cost of debt model used to test H5. The mean (median) value of COD is (0.027), which is much lower than those reported by Francis et al. (2005) for a number of international companies from 42 countries (0.113 and for mean and median, respectively). As shown by the correlation matrix, no apparent relationship exists between COD and the independent variables. The regression results are reported in Table 6. Contrary to our prediction, the model is not statistically significant and the explanatory power is very low. Each of the coefficients on disclosure variables is insignificant, with some of them even negative. In short, we found no evidence to support the effect of disclosure on the companies cost of debt, and thus H5 is not supported. Among the control variables, only the coefficient on ROA is negatively significant, indicating profitable firms are more likely to attract lower cost of debt. Firm size and leverage level were not predictive of cost of debt. The lack of expected association between cost of debt and the test variables can be argued on the basis of two aspects. First, there may be independent variables that we did not control for due to data unavailability (for example a firm s need for external financing). Second, the debt market in China is still underdeveloped, especially the bond market. Debt is not only an abundant source of funds for the growth of China s economy, but also an important part of the fiscal policy used by the central government to exercise macro-control. A company s debt obligations are not a complete reflection of its financing needs guided by the market mechanism, and as such, the cost of debt financing is not fully responsive to the environment variables. In this regard, the Chinese government needs to introduce more rules to promote the development of a unified debt market, increase transparency of debt issuance, adjust the term structure of debts, and gradually introduce market-based pricing in debt issuance.