Types of guarantees and their relation to external auditing：Evidence from the Chinese bond market

Abstract

Guarantee conditions and external auditing are both key parts of debt contracts. Understanding the relationship between guarantee conditions and external auditing is crucial for the improvement of debt contract efficiency and maximization of enterprise value. With non-listed enterprises in the Chinese bond market from 2008 to 2013 as the study sample, this paper investigates the relationship between different types of guarantee and external auditing. The empirical result indicates that: a related-party guarantee is significantly positively correlated with high-quality external auditing, while a local-government guarantee (in which the guarantor and bond issuer belong to the same local government) is significantly negatively correlated with high-quality external auditing. The research also shows that high-quality external auditing is significantly negatively related to bond issuance cost only if companies use a related-party guarantee. The reason is that a local-government guarantee implies the credit and security of local government, which reduces debt agency conflict and the demand for high-quality external auditing, while a related-party guarantee does not reduce information asymmetry, but increases the credit risk of bonds. This risk can be recognized by investors in the Chinese bond market which is dominated by institutional investors, thus causing stronger demand for high-quality external auditing. The result shows that a guarantee substitutes for external auditing when the guarantee is positive, otherwise external auditing is a supplement to a guarantee.

Keywords:

• bond issuance cost
• bond market
• debt contracts
• debt guarantee
• external auditing

1. Introduction

In recent years, the Chinese bond market has developed rapidly. By the end of November 2012, the total value of the bond market stock had reached RMB 25.6 trillion, exceeding the total market value of the Shanghai and Shenzhen stock markets.¹ The increasing importance of the bond market in the financial system gives rise to a key research subject in theory and practice on how to build a sound investor protection mechanism to maintain the stability and development of the bond market.

Information asymmetry in the capital market poses potential risks to the market mechanism and to investors, which raises two basic problems, namely an information problem and an agency problem (Wei, Chen, & Li, 2007). The information problem stems from ex ante information asymmetry, which results from information disparity and conflict of interest between enterprise insiders and outside investors. Therefore, ‘adverse selection’ emerges, which may lead to market failure (Akerlof, 1970). Under such circumstances, a guarantee may help the creditor to improve its ability to evaluate expected returns and serve as a signal to relieve adverse selection in the credit market (Chan & Thakor, 1987; Stiglitz & Weiss, 1981). Likewise, external auditing may also reduce information asymmetry between outside investors and corporate insiders and convey to the investors a positive signal of efficient governance and reliable financial statements (Datar, Feltham, & Hughes, 1991; Fan & Wong, 2005), thus avoiding adverse selection by the creditors. The agency problem originates from ex post information asymmetry between the creditors and the corporate managers, which may lead to entrenchment by the latter. Jensen and Meckling, (1976, pp. 355–356) indicate that there is research interest in investigating the actions of creditors as the owners of collateral. In addition, audit quality appears as a relevant monitoring device to regulate agency conflicts between shareholders and debtholders (Piot, 2001).

Considering the significant role of guarantee and auditing, Chinese regulators take the use of a guarantee and the application of auditing as basic system requirements in the bond market. The National Development and Reform Commission (hereinafter referred to as ‘NDRC’) issued in 2004 a Circular on Further Improving and Strengthening Control over Enterprise Bonds, specifying firstly that an independent legal person must be engaged to guarantee the bond issuance of enterprises and secondly that financial statements of the most recent three years furnished by the issuers should be audited by a qualified accounting firm.² The document Circular on Matters of Standardizing Listed Trading of Enterprise Bonds in the Stock Exchange promulgated by the China Securities Regulatory Commission (CSRC) in 1996 also stipulates that enterprise bonds applying for listing must be guaranteed and that audited financial statements for the most recent three years of the issuers should be included with the listing announcement. The decision must be made on whether to provide a guarantee prior to issuing bonds and the shareholders’ meeting must make a resolution concerning the guarantee matters (Sui & Ming, 2012).

Despite the wide use of guarantees and external auditing in the practice of issuing bonds in China, there is little in-depth theoretical study about the relationship between guarantee conditions and external auditing in the bond market. Based on an American Bar Foundation compendium entitled Commentaries on Indentures, Smith and Warner (1979) investigate the various kinds of bond provisions that are included in actual debt contracts. They reveal that the guarantee and auditing are the crucial parts of a debt contract that can effectively alleviate agency conflict between the bond investors and shareholders. They also point out that enterprises are not able to satisfy all the terms of debt contracts due to high costs. Therefore, optimal debt contract terms may be arranged to maximize corporate value. However, they did not clearly explain how to design the debt contract terms but considered it necessary to research substitutive and supplementary relationships between various terms to determine which form of debt contract to adopt. Extant researches focus on the respective role of guarantee and external auditing in alleviating agency conflict between the shareholders and creditors. For example, Hsueh and Kidwell (1988) analyse the Texas School Bond Guarantee programme and its impact on the borrowing cost of school districts and other municipalities within the state. Quigley and Rubinfeld (1991) research the market for private guarantees for US municipal debt and provide empirical evidence on the effects of bond insurance upon the pricing of new issues. Using a large sample of US firms, Mansi, Maxwell, and Miller (2004) find a negative relation between auditing quality and tenure and the return investors require on corporate bonds. Kim, Simunic, Stein, and Yi (2011) take advantage of the institutional setting in Korea to assess the value of external auditing per se and the possible value differences between Big Four and non-Big Four audits in the context of the pricing of private debt such as bank loans. However, the relationship between guarantee and auditing and their interaction on bond financing have remained uninvestigated. Moreover, due to the lagging development of the Chinese bond market, very few Chinese researchers have paid attention to the role of guarantees in the bond market. In the primary period of the bond market, bond guarantee is compulsory and dominated by bank guarantee, not leaving room for research. With the further development of the Chinese bond market, bond guarantees have evolved to voluntary guarantees provided by the enterprise under market conditions, leading to a pattern of diversified guarantees. Some of them are asset backed with equity, land use rights and real estate as collaterals and some provide with third-party guarantees including relate-party guarantee, mutual guarantee and local-government guarantee. In legal form, such guarantees comply with laws and regulations and bond issuance requirements; however, they may have different guarantee effects in practice. Collateral enables the creditors to make up for losses by liquidating mortgage assets in the case of bond default and therefore is supposed to have a positive effect. A local-government guarantee implies the supply of credit and the provision of insurance from the local government and thus should have the strongest positive effect. Related-party guarantee and mutual guarantee may easily result in capital chain break of connected parties, rendering very high credit risk, and are considered to have negative effect rather than positive effect.³ Differences among the above various guarantees provide an interesting opportunity for the further study of guarantees. It is also interesting to investigate the way in which various guarantees with opposite guarantee effects relate to external auditing. A clear identification of the effect of various guarantees and an understanding of the relationship between various guarantees and external auditing may contribute to the reduction of debt contract cost and an increase of debt contract efficiency.

Based on the above analysis, and with the Chinese bond market as the study object, this paper seeks to answer the following questions: do various guarantee terms of debt contracts have positive effects in the bond market? What is the relationship between guarantees and external auditing? Can external auditing serve as a supplementary mechanism to reduce debt agency cost in the absence of a positive guarantee? The evidence will aid understanding of the role of debt guarantee and external auditing in the bond market and will enrich research on guarantees and auditing. In terms of practical meaning, on the one hand, testing the effects of various guarantees in the bond market helps the investors to identify credit risk so as to make a correct investment decision. It also helps the regulators to formulate regulation measures for standardizing the guarantee behaviour of enterprises. On the other hand, the issuing firm could be forced to bear high bond issuance cost or even withdraw from the bond market in the absence of positive guarantee. If external auditing is a supplement to guarantee, external auditing will serve as a supplementary mechanism to alleviate information asymmetry between both parties and reduce investment risk, financing the cost of enterprises and the transaction cost of the whole bond market, thus promoting the healthy development of the bond market.

Compared with previous studies, this paper makes the following contributions. First, it covers a gap in the existing literature on protecting the rights of creditors (Sloan, 2001; Wei et al., 2007) using a guarantee and external auditing. Second, it categorizes guarantees in the bond market in detail and tests their respective roles in the bond market. Although prior literature has investigated incentives for, and influences on the corporate value of guarantees by listed companies (Rao, Zhang, & Peng, 2008; Wan & Wei, 2009; Wang & Chen, 2007; Zheng, Fan, & Zhu, 2007), matters concerning the bond market have not been mentioned. Li (2006) has analysed theoretically the relationship between the compulsory guarantee of Chinese bond market and the investor’s position, but no systematic empirical test has been undertaken of the effects of various guarantees. This paper analyses institutional and economic roots for the generation of various guarantees from the perspective of institutional changes in the Chinese bond market, and systematically compares their effects by investigating enterprise bonds issued from 2008–2013. Third, it proves tunnelling by controlling shareholders in the Chinese bond market. Previous literature provides empirical evidence on tunnelling by controlling shareholders in the Chinese stock market (Huang & Yin, 2008; Wei, Huang, & Cheng, 2013; Zhang & Li, 2010). This paper finds that a related-party guarantee is positively correlated with high-quality external auditing, indicating that the controlling shareholder of the bond issuer is motivated to provide a related-party guarantee to help the issuer obtain debt funds, which increases the resources available to expropriate. This finding also serves as empirical evidence on tunnelling by controlling shareholders in the Chinese bond market. Fourth, using a unique sample, it explores the influence of agency conflict between the bondholders and shareholders on the demand for high-quality external auditing, further supporting agency theory and auditing demand theory. The empirical study by Chow (1982) indicates that debt agency cost imposes significant influence on auditing demand, but his indicators for measuring the debt agency cost are too rough. Meanwhile, the influences of equity agency costs are not well controlled, thereby undermining the robustness of the study. This paper takes the enterprise bonds of non-listed companies as its sample and lowers the influence of agency conflicts between shareholders and managers as well as controlling shareholders and minority shareholders on auditing demand. In this way, we can better identify the effects of debt agency cost on the demand for high-quality external auditing. Finally, the empirical results of this paper prove that there is a substitutive relation between a positive guarantee and external auditing, which means that a positive guarantee decreases demand for high-quality external auditing. Meanwhile there is a supplementary relation between a negative guarantee⁴ and high-quality external auditing, meaning that a negative guarantee strengthens the demand for external auditing.

The remainder of this paper is structured as follows. Section 2 reviews the literature. Section 3 describes the institutional background of the Chinese bond market. Section 4 develops our hypotheses. Section 5 describes the research design adopted for this study. Section 6 reports our empirical results and performs robustness tests. Section 7 concludes the paper.

2. Literature review

Clearly, all actors in the municipal bond market do not possess the same information about the creditworthiness of issuers (Quigley & Rubinfeld, 1991). Information asymmetry between enterprises and creditors may result in entrenchment of the creditors by managers or shareholders (Sun, Li, & Wang, 2006). There are many economic mechanisms to mitigate the undesirable influence of information asymmetry. Providing a guarantee is one of the most effective ways (Akerlof, 1970). Where the creditors expect that the debtors may infringe their interests, the creditors have strong incentives to limit the behaviour of managers to devalue the bonds by employing various contract terms including pledges and guarantees (Jensen & Meckling, 1976). Bond sellers could be encouraged to issue costly signals of their own about the quality of bonds offered for sale; in equilibrium, prospective bond buyers could use these signals to discriminate accurately among bonds of differing quality (Ross, 1977). Quigley and Rubinfeld (1991) believes that it seems reasonable to expect that the issuers of bonds know the most about their own capacities to repay debt obligations while potential holders of the obligations would have to invest even more to obtain equivalent information. Insurance offerings provide information in this world of asymmetric information. Quigley and Rubinfeld (1991, p. 32) comment that this characterization of information suggests several rational responses to the failure inherent in the market for ‘lemons’. Hence, guarantee is an important contract mechanism to mitigate debt agency cost (Chen, Lobo, Wang, & Yu, 2012).

External auditing is another contract mechanism to alleviate information asymmetry in the bond market and reduce debt agency cost. Jensen and Meckling (1976) argue that independent auditors would be engaged by management to testify to the accuracy and correctness of accounting reports and to alter the opportunity the owner-manager has for capturing non pecuniary benefits. Raman and Wilson (1994) point out that high-quality auditing effectively reduces default risk of the bonds. Kim, Simunic, Stein, and Yi (2011) consider that retaining a Big Four auditor enhances the credibility of financial statements and reduces ex ante information uncertainty and ex post debt monitoring costs. Previous studies have proved with empirical evidence the role of external auditing in reducing debt agency cost. It has been revealed in researches of Mansi et al. (2004), Pittman and Fortin (2004) that the debt cost of enterprises engaging hiring a Big Six auditor is lower than those not doing so. Minnis (2011) compares the influences of three attestation methods (provided by auditors) of financial statements: review, compilation and auditing on debt pricing (interest rate) of the enterprises. Their results show that auditing can significantly improve the usefulness of the financial statements when compared with review and compilation. This explains the lower debt cost of audited enterprises. Chen and Li (2013) examine the influence of external auditing on bond credit rating by taking a sample of corporate bonds issued by listed companies in China. The results reveal that external auditing can improve the credit rating of corporate bonds.

Although guarantees and auditing are important debt contract mechanisms, not all enterprises provide the same guarantees and choose auditors of the same size. Zhang, Feng, and Zhang (2011) studied the situation of 3479 listed companies in China hiring external auditors during 2003–2008 and concluded that 27.19% firms did hire ‘Big Four’ or domestic ‘Big Ten’ and 72.81% did not. Chen et al. (2012) studied 5358 listed companies in China during 2001–2006 and found that 78.2% of them adopted collateral while 21.8% had no collateral. The reason for differences between Chinese enterprises in guarantee and external auditing possibly lies in the high cost of debt contracts, which makes enterprises unable to provide all debt contract terms. Various terms have different costs and are mutually replaceable, resulting in significant differences in debt contracts of various enterprises. Smith and Warner (1979) argue that optimal debt contracts can only be arranged by fully understanding the relationship between various contract terms to maximize corporate values.

Some scholars have studied the relation between accounting and other terms in debt contracts. Kim, Tsui, and Yi (2011) studied the influence of voluntarily adopting International Financial Reporting Standards (IFRS) on the price term (interest rate) and non-price terms (loan size, loan maturity, collateral guarantee and restrictive terms) in bank loan contracts. The results revealed that the banks tended to offer lower interest rates, larger loans, longer maturity and less restrictive terms for enterprises adopting IFRS. For the purpose of this current research, that evidence is interpreted as indicating that there is a substitutive relationship between accounting terms and other terms in debt contracts. Beatty, Weber, and Yu (2008) reveal that income escalators in net worth covenants are positively correlated with accounting conservatism, and the creditors may adopt accounting conservatism and income escalators to reduce debt agency cost. Nikolaev (2010) finds that firms with more extensive use of terms in their public debt contracts exhibit higher levels of timely recognition of economic losses in accounting earnings. For the purpose of this current research, that evidence is interpreted as indicating that there is supplementary relation between accounting terms and other terms in debt contracts. Chen et al. (2012) further the study of collateral and accounting conservatism and find a negative relation between accounting conservatism and the use of collateral, which shows that collateral and accounting conservatism are substitutive to each other in decreasing debt agency cost. However, they also find that the lender uses both collateral and accounting conservatism to manage risk when the borrower’s observed credit quality is low or when the borrower has a small fraction of total assets that are tangible. Their findings may be interpreted as suggesting that accounting terms can substitute or complement other terms in different debt contracts.

Although the foregoing literature offers an in-depth study of the role of either guarantees or external auditing in debt contracting, it does not cover guarantees and external auditing at the same time, let alone their relation to debt contract. Researchers have examined the correlation of accounting with interest rate, bond period and other terms in debt contract; however, the correlation between guarantee types and auditing remains unstudied. More importantly, diversified guarantee types in the Chinese bond market may have opposite effects, causing a varied relation between types of guarantees and external auditing. This paper first distinguishes and examines the effects of various guarantee types by using a unique sample from the Chinese bond market. On this basis it further researches the substitutive or supplementary relationship between various guarantees and external auditing and the influence of their interaction on bond issuance cost, offering new empirical evidence for agency cost theory and debt contract theory.

3. Institutional background

Bonds issued by enterprises in China are divided into enterprise bonds, corporate bonds, convertible bonds, separate convertible bonds, medium term notes and short-term financing bills. Various bonds differ greatly. Convertible bonds and separate convertible bonds have very short terms and usually can be converted into equity stocks within half a year. They include call options (Chen & Li, 2014), which make them differ from pure bonds and unsuitable for researching debt agency problem. Short-term financing bills refer to bonds whose principal and interest shall be paid within one year. Medium term notes are also issued depending on credit and therefore, short-term financing bills and medium term notes cannot be used for studying a guarantee. Corporate bonds are issued by listed companies. Prior research indicates that agency conflicts between shareholders and managers as well as controlling shareholders and minority shareholders have a significant impact on the demand for high-quality external auditing (Tang, 2011; Wang, 2009). It is hard to tell whether the demand for high-quality external auditing stems from manger-shareholder agency conflict or shareholder–bondholder agency conflict in corporate bonds. Besides, corporate bonds, convertible bonds and separate convertible bonds can only circulate in stock exchange markets while medium term notes and short-term financing bills can only circulate in the interbank bond market. The segmentation of Chinese bond markets and their different systems of regulation limit the use of the above-mentioned bonds to fully reflect current conditions of guarantees and auditing in the bond market. Enterprise bonds can circulate in stock exchange markets and the interbank bond market with a term of 5–15 years in general. Enterprise bonds are issued by non-listed firms. Compared with listed firms, non-listed firms have less ownership dispersion and less owner-manager separation (Hope, Langli & Thomas, 2012), so non-listed firms’ agency conflicts between shareholders and managers as well as controlling shareholders and minority shareholders are trivial. The agency cost of non-listed firms is less than that of listed firms (Chaney, Jeter, & Shivakumar, 2004; Hope, Thomas & Vyas, 2014). The lower agency costs of equity of non-listed firms help them to specialize in agency costs of debt and debt contracts. Therefore, enterprise bonds by non-listed firms provide a unique sample for studying the relationship between various guarantee terms and external auditing in debt contracts.

In the early phases of development of the China enterprise bond market, the government enforced the provision of a guarantee before bond issuance by the enterprises, for the purpose of protecting individual investors and maintaining bond market stability (Li, 2006). Bonds are mostly guaranteed by non-commercial banks and state-owned commercial banks. Rating agencies rated bonds mainly according to credit rating of the guarantee banks so that almost all bonds are rated as AAA (Li, 2006; Shi & Yang, 2011). Excessive guarantee by state-owned banks for the enterprise bond market leads to a transfer of the credit risk of the bond market to the banking system, thus increasing the risk of the financial system and putting bank assets in danger (China Banking Regulatory Commission, 2007). Therefore China Banking Regulatory Commission (CBRC) promulgated on October 12, 2007 a document: Opinions on Effective Prevention of Enterprise Bond Guarantee Risk (hereinafter referred to as the Opinions 2007), requiring the banks to stop granting a guarantee for enterprise bonds.

After the implementation of the Opinions 2007, banks exited the bond guarantee market. Collateral and third-party guarantees are now the major guarantee types in the interbank bond market. For collateral, companies take equity, land use right and real estate as assets for security. Among third-party guarantees, enterprises are the main guarantee force in the interbank bond market.⁵ According to relations between the guarantors and bond issuers, guarantees are divided into related-party guarantee (related-party guarantee means that bond issuers are guaranteed by their controlling shareholders), mutual guarantee (mutual guarantee means that the bond issuer and their guarantor provide a guarantee for each other) and local-government guarantee (in this paper the term ‘local-government guarantee’ means that the bond issuer and their guarantor are both state-owned enterprises that are affiliated to the same local government) insured by local government. Although these guarantees comply in form with the regulations for bond issuance, they do not have the same guarantee effects. If the controlling shareholders of an issuer provide a related-party guarantee for the issuer, there is a strong correlation between the guarantor and the issuer and the guarantees play a weak role (Shi & Yang, 2011). Some enterprises choose a mutual guarantee for the purpose of bond sales rather than risk reduction. A mutual guarantee increases the explicit and implicit debts of the issuer, resulting in a higher default risk for the enterprise (Shi & Yang, 2011). A local-government guarantee virtually implies assurance by the local government. The implicit guarantee by the local government helps ease the worries of the creditors about the default risk of the state-owned firms, and the creditors will not worry about the solvency of state-owned firms as much as that of private firms (Chen, Chen, Lobo, & Wang, 2010).

Another distinct feature of the Chinese bond market is that institutional investors are the main players in bond market. Institutional investors include commercial banks, trust & investment companies, securities firms, fund corporations, insurance companies, finance companies of enterprise groups and other similar entities. According to the statistics of China Central Depository & Clearing Co. Ltd. (CCDC), institutional investors held 98.00% of the total amount of bond custody in July 2014, among which commercial banks held 63.66%.⁶

4. Theoretical analysis and research hypotheses

4.1. Theoretical analysis

The research result reported by Chen et al. (2012) indicates that a supplementary or substitutive relation between various terms of debt contracts depends on the credit quality of the debtors and the debt default risk faced by the creditors. If the credit quality is low and the debt default risk is high, it is required that the debtors should provide more terms in debt contracts to protect the creditors’ rights, thus presenting supplementary relation between various contract terms. When the debt default risk is low, the creditors will take one of the terms as a substitute for other contract terms for the credit risk control target (Chen et al., 2012), which means various terms may show a substitutive relation.

In the light of its information function, external auditing helps to alleviate ex ante information asymmetry faced by creditors and reduce the costs of ex post supervision and re-bargaining, hence increasing debt contract efficiency (Kim, Simunic, Stein, & Yi, 2011). However, the role of external auditing as a contract term⁷ is closely related to effects of other terms in the debt contracts. When other terms are able to effectively alleviate agency conflicts between the debtors and creditors and the debt default risk faced by the creditors, the role of external auditing is less significant, the demands for external auditing by the creditors will decrease accordingly. On the other hand, external auditing can serve as a substitute for other terms to effectively address debt agency cost. For example, using US data for 2003 to 2006, Gul and Goodwin (2010) found that short-term debt and credit rating quality are negatively related to audit fees for firms rated by Standard & Poor’s. Lou and Vasvari (2013) demonstrate that reputable auditors help corporate bond issuers obtain lower bond yields and firms that hire reputable auditors obtain longer term bonds. This shows that such contract terms as debt maturity and price are in substitutive relations with external auditing in terms of reducing debt agency cost, thus they attest the ‘Costly Contracting Hypothesis’ put forward by Smith and Warner (1979). Similarly, due to high costs of guarantee and auditing in the bond market,⁸ the demand for high-quality external auditing could be reduced for economical purpose when guarantee alone can effectively reduce debt default risk faced by the creditors. However, the ‘Costly Contracting Hypothesis’ fails to hold when a negative guarantee provided by the debtors increases the debt default risk rather than reducing it. Under such circumstance, the creditors still need high-quality external auditing to enhance the reliability of the financial statements and provide credit risk information of the debtors (Kim, Simunic, Stein, & Yi, 2011) for risk analysis and investment decision-making. From the viewpoint of suppliers, if the investors can identify a negative guarantee, the debtors also need external auditing to convince the creditors of a good financial condition and corporate governance of the enterprises to ensure successful issuance of the bonds and reduce bond issuance cost. Hence, when the guarantee is negative, external auditing may serve as a supplement to alleviate agency conflict between the creditors and debtors.

The precondition for various guarantees to take positive or negative effects is the capacity of the creditors to identify the credit risk in the credit market. If the investors can recognise the risk that their interests will be infringed, distinguish and keep away from defaulting companies, securities issuers and purchasers in the capital market can widely use private contracts such as ‘intermediary guarantee’ to improve efficiency in resource allocation (Ji & Cao, 2008). Otherwise, credit risk will not be identified by the creditors even though the guarantee is negative. It will become a ‘true false-news’ only to attain the objective of signalling to the capital market (Tang, 2011)⁹. The main participants in the bond market are institutional investors, who understand income manipulation better than individual investors (Balsam, Bartov, & Marquardt, 2002), use information more correctly than individual investors (Cohen, Gompers, & Vuolteenaho, 2002) and have a stronger ability in information-interpretation (Wang, Liu, & Wu, 2009). Compared with individual investors, institutional investors have a large investment and their scale economy and diversification of investment allow them to benefit from cost advantage in information analysis (Gao, Zhou, & Wang, 2011). Such cost advantage motivates them to collect and analyse information about the issuers. On the other hand, a high professional calibre equips the institutional investors with strong risk awareness and risk identification ability to distinguish between the effects of various guarantees and the default risk of various bonds.

4.2. Research hypotheses

Among enterprise guarantees, a related-party guarantee provided by the controlling shareholders for bond issuers is a typical related-party transaction. A related-party transaction is an important means of tunnelling by controlling shareholders (Johnson, La Porta, Lopez-De-Silanes, & Shleifer, 2000). To help a company obtain debt funds that favour expropriating more resources, the controlling shareholder of the bond issuer could be stimulated to provide a related-party guarantee for the company. In Asia, the controlling shareholder is free to increase the firm’s leverage and build up the company’s resources to expropriate (Bai, Lin, Wang, & Wu, 2013). Du and Dai (2005) find that controlling shareholders tend to increase leverage to acquire more resources without diluting their ownership dominance. Bai et al. (2013) find a positive and significant relationship between expropriation and debt, which provides empirical evidence that controlling shareholders exploit minority shareholders through debt. The expropriation of controlling shareholders leads to discounts in the stock market value of the firm (Lins, 2003; Wei et al., 2013; Zerni, Kallunki, & Nilsson, 2010), which also impairs the interest of creditors. The expropriation of controlling shareholders essentially transfers to the creditor of the company once the company falls in bankruptcy (Zheng, Fan, & Zhu, 2007). Consequently, if the controlling shareholders do have a motive to ‘tunnel’ in bond issues by subsidiaries and such intention has been identified by institutional investors, a related-party guarantee may work adversely rather than play a positive role. In such cases, investors will request high-quality auditing to intensify the corporate governance of bond issuers to prevent expropriation of controlling shareholders. In this way, a supplementary relation between related-party guarantee and external auditing is established.

For creditors, a mutual guarantee provided by two enterprises for each other implies high credit risk. The greatest risk of a mutual guarantee is that the secured parties maliciously borrow money and misuse bank funds to make the total amount of guarantee exceed their solvency. Once the lending banks file an action in the court, all companies providing a guarantee shall assume joint liability and face potential litigation risk (Lv and Wang, 2007). Chen and Cao (2011) conduct a questionnaire survey on SMEs (small and medium sized enterprises) in Zhejiang to find that when a business owner absconds to avoid debt payment, due to the impact of mutual guarantee, those enterprises providing the guarantee are implicated, which amplifies the adverse influence. Therefore, mutual guarantees increase the default risk of enterprises, seriously damage creditors’ rights, and lead to an increase of financial risk in the credit market and welfare loss instead of giving full play to the information and governance function of the guarantee. When bond issuers provide a mutual guarantee, institutional investors discern implicit credit risk and the agency conflict between bondholders and bond issuing firms increases. Firms will be forced to hire high-quality external auditors to meet the bondholders’ need of more high-quality accounting information in analysing the financial condition and solvency of firms. In this way, a supplementary relation could be established between mutual guarantee and high-quality external auditing.

Under a local-government guarantee, bond issuers and bond guarantors are both state-owned enterprises in the same region. As the ultimate controllers to the state-owned enterprises in the same region are the local governments, governments can play the role of coordinator and require some state-owned enterprises to provide a non-reciprocal guarantee for other state-owned enterprises in the same region. These guarantees, supported by government, improve bond ratings¹⁰ (Shi & Yang, 2011). Such actions of local governments in terms of enterprise bond guarantee are conveying to the bond market a signal that since the local government intervenes in enterprise guarantee, it cannot escape when the issuers are trapped in a financial crisis and it has a moral obligation to assist such enterprises in the case of insolvency. A local-government guarantee for enterprise bonds implies credit and insurance by the local governments. This ‘implicit guarantee’ from local governments can significantly reduce the default risk of bonds. With a government insured guarantee, demand for other contract mechanisms including external auditing by the investors may decrease significantly. Thus, it is possible that there is a substitutive relationship between local-government guarantee and high-quality external auditing.

Collateral also reduces the default risk of bonds to a certain extent. When the borrowers have no distinct characteristics to signal their risks to the creditors, they use collateral to indicate their quality (Booth & Booth, 2006). Helmut (1994) proves that the possibility of renegotiation implies that default will not always be penalized by bankruptcy and both parties to the loan realize this. Helmut (1994, p. 73) indicates that, knowing that there is a chance of debt forgiveness, the borrower may falsely claim that the debt exceeds the investment’s return and that he is forced to default. This motive for cheating is weakened when collateral has been posted. The higher the degree of collateralization, the more inclined is the creditor to believe that the project return actually is low when he observes default. Consequently, he finds the option of taking over the project less profitable in comparison to forgiving a portion of the debt. In this way, collateral may reduce the expected cost of bankruptcy. Booth and Booth (2006) find that collateral has wider use when default risk increases. It shows that collateral can serve as a signalling mechanism to alleviate adverse selection due to the inability of the creditors to understand the risk of the debtors. Thus, after enterprises provide collateral, bond investors may have less demand for external auditing, which means that there is a possible substitutive relation between collateral and external auditing. However, the effect of collateral is largely dependent upon the degree of realization of security rights to pledged assets by the creditors in case of bankruptcy. Since the formalities of collateral in China are cumbersome without an effective mechanism for the execution of a guarantee, collateral is of low efficiency and high transaction cost. In proceedings on guarantee, it is very common that those who win the case suffer financial loss (Liu, 2009). We can see that collateral alone is unable to eliminate credit risk (Yang & Qian, 2008). Hence, the effect of collateral is undermined to some extent. It is possible that the effects of collateral are weaker than those of a local-government guarantee, which makes the substitutive relation between collateral and high-quality external auditing non-significant.

In conclusion, among the four types of guarantee described above, collateral and local-government guarantee are positive guarantees, while related-party guarantee and mutual guarantee are negative guarantees due to their relatively high credit risk. Among positive guarantees, a local-government guarantee implies credit and insurance of the local governments and thus has the strongest guarantee effects. Collateral has a certain effect, but its guarantee effect may be weaker than that of a local-government guarantee. When bond issuers adopt a negative guarantee, there is a supplementary relation between the guarantee and high-quality external auditing. When bond issuers adopt a positive guarantee, there is a substitutive relation between guarantee and high-quality external auditing. The more effective is the positive guarantee, the more substitutive will be the relation between the guarantee and high-quality external auditing. Based on the above analysis, this paper proposes hypotheses as follows:

• Hypothesis 1-1: Negative guarantees are positively correlated with the high-quality external auditing (supplementary relationship).

• Hypothesis 1-2: Positive guarantees are negatively correlated with the high-quality external auditing (substitutive relationship).

Other than testing the substitutive and supplementary relationships between guarantee and high-quality external auditing, this paper also examines the relationship among guarantee, high-quality external auditing and bond issuance cost. As a key approach of credit enhancement, a guarantee helps to effectively lower bond default risk (Thakor, 1982). Banks are concerned during the rating of credit granting about whether the borrowing enterprise provides a guarantee. The existence of a reliable guarantee means high protection for the creditors. It is the same with enterprise bonds (Li & Cao, 2009). Credit enhancement (guarantee) of bonds improves credit rating and helps enterprises with low credit rating to finance via the bond market and save financing cost (Zhu & Duan, 2012). Chinese and foreign research has also revealed that a guarantee can effectively reduce the bond financing cost. Kidwell, Sorensen, and Wachowicz (1987) study the correlation between bond insurance in the American bond market and the bond issuance cost to find that the interest cost savings from purchasing insurance exceeds the cost of the insurance premium. Furthermore, the net benefit to the issuer increases as the underlying credit quality of the bond declines. Hsueh and Kidwell (1988) examine the impact of a state bond guarantee programme of municipal bonds with Texas schools as an example, and discover that Texas school districts achieve interest cost savings as a result of the guarantee programme.

External auditing also helps to reduce debt cost, and its pricing function is closely related to the degree of information asymmetry and credit risk in the credit market. According to Raman and Wilson (1994), in addition to selecting ‘Big Eight’ auditors which signals high-quality auditing, the standardization of government auditing procurement can also transmit the signal of high-quality government auditing, thus reducing bond financing cost. By taking ‘Big Six’ auditors as indicators for measuring high-quality auditing, Mansi et al. (2004) find that high-quality auditing is significantly and negatively related to debt cost and that the relation between high-quality auditing and debt cost is most pronounced in firms with debt that is non-investment grade. Pittman and Fortin (2004) find that for newly public companies that experience serious information asymmetry in their early public years, choosing a ‘Big Six’ auditor can lower firms’ interest rates significantly. However, with the development of the firms, information asymmetry between firms and lenders is gradually reduced and the influence of the ‘Big Six’ is weakened. Prior research by Blackwell, Noland, and Winters (1998) and Dharan (1992) also indicate that since small firms have information asymmetry, they use external auditing to convey high-quality signals to the credit market. However, with the increase of the scale and the growth of firms’ reputation, the role of auditing in reducing corporate debt cost diminishes.

According to this prior research, the greater the information asymmetry between the creditors and debtors, the higher the credit risk and the more significant the role of external auditing in cutting debt cost, and vice versa. Likewise, if issuers offer negative guarantees such as related-party guarantee and mutual guarantee, the correlation between the guarantors and issuers is very strong, with poor guarantee effect (Shi & Yang, 2011), and the two guarantees are unable to reduce the credit risk of the issuers. Consequently, bond investors still need external auditing as a supplement to alleviate debt agency cost, and external auditing may significantly reduce bond issuance cost. On the other hand, if the issuers provide positive guarantees such as local-government guarantee and collateral, information asymmetry between the investors and issuers will be effectively alleviated and default risk and bond issuance cost will be reduced. In consideration of the substitutive relation between positive guarantee and external auditing, the effect of external auditing on reducing bond issuance cost may not be very great. On the base of the above analysis, this paper proposes hypotheses as follows:

• Hypothesis 2-1: For enterprises providing negative guarantees, high-quality external auditing can reduce bond issuance cost significantly.

• Hypothesis 2-2: For enterprises providing positive guarantees, high-quality external auditing has no significant influence on bond issuance cost.

5. Research design

5.1. Sample selection

Enterprise bonds include the ‘urban construction bond’ issued by local government’s financing platforms and the enterprise bond issued by common enterprises. Since the ‘urban construction bond’ is obviously related to and supported during issuance by local governments, the role of guarantee and external auditing is greatly diminished. For this reason, this paper chooses enterprise bonds issued by common enterprises as the population. Subject to Notice on Matters of Further Enhancing the Management of Enterprise Bond Risk Prevention issued by NDRC in 2012, ‘bond issuance by local state-owned enterprises for infrastructure investment including construction of tourism infrastructure, logistics park, development area and industrial parks well as water conservation and roads building, is in principle under the management of government investment and financing platform due to its connection with government investment plan and various policies’. It is on the base of this provision that this paper defines ‘urban construction bond’ and distinguishes it from common enterprise bonds. During 2004–2013,¹¹ there were 1488 enterprise bonds in total, listed in the stock exchange market or the interbank bond market, including 307 enterprise bonds issued by common enterprises other than ‘urban construction bonds’ and bonds of financial insurance enterprises. A mandatory guarantee was imposed for enterprise bonds issued from 2004–2007 resulting from government regulation rather than market selection¹² and thus does not represent the demand of the bond market for a guarantee. Therefore, the population of enterprise bonds guaranteed by banks is removed in subsequent multiple regression analysis to ensure study robustness. After excluding the enterprise bonds guaranteed by banks and some other minor guarantee types, this study period covers 2008–2013 with 227 bonds.¹³ All continuous variables in the sample are winsorized at 1% and 99% points. Such variables as guarantee, external auditing and financial indicators of the issuers are manually collected and compiled from bond issuance documents including enterprise prospectus, auditing reports and credit rating reports disclosed on Chinamoney.com.cn and Chinabond.com.cn.

5.2. Research model and variable definitions

This paper builds the following logistic regression models to test Hypothesis 1 and Hypothesis 2:(1) (2)

Model (1) tests the influence of a guarantee on demand for high-quality external auditing, in which Guarantees represent various guarantees; Auditing represents high-quality external auditing. In the Chinese bond market, external auditing is not voluntary but compulsory, however, the bond issuing firms have a right to choose different accounting firms. When laws require a firm to supply external auditing, the relation between agency conflict and demand for external auditing develops into the firm’s motive to hire high-quality external auditing (Wang, Chen, & Yu, 2006). Therefore, the paper takes the size of the accounting firm hired by the bond issuer as the measure of high-quality external auditing.¹⁴ As to the size of accounting firms, rankings in the Information of Top 100 Auditors in Comprehensive Evaluation issued by the Chinese Institute of Certified Public Accountants (CICPA) for 2004–2012 are adopted as criteria for determining whether an auditor is a Chinese ‘Big Ten’ auditor. Chinese ‘Big Ten’ means that bond investors will have a stronger demand for external auditing.¹⁵ Independent variables include related-party guarantee, mutual guarantee, local-government guarantee and collateral during bond issuance for testing the effects of various guarantees on the choice of high-quality external auditing. As the research topic in this paper is on unlisted companies¹⁶ issuing enterprise bonds in the bond market, when testing the influence of a guarantee on the choice of high-quality external auditing, the selection of control variables has borrowed from overseas studies including Mansi et al. (2004); Brandon, Crabtree, and Maher (2004), and Chinese studies including Lei, Li, and Wang (2009) and Pan (2010). Among them, Return-on-Assets is the average return on assets (the ratio of earnings divided by the total assets) three years before issuance; Leverage is asset-liability ratio one year before issuance; Current-Ratio is the current ratio (current assets divided by current liabilities) one year before issuance; Times-Interest-Earned is the times interest earned ratio (income before interest and tax expense divided by interest expense) one year before issuance; Ln(Sales), representing enterprise size, is the natural logarithm of sales one year before issuance; State-Owned Enterprise is a dummy variable reflecting government ownership, it equals 1 if the firm is a state-owned enterprise and 0 otherwise; Year is the year variables (five annual dummy variables are set in the model with 2008 as the reference year); Industry is industry variables (issuers are divided into 12 industries excluding financial and insurance industries) and 11 industry dummy variables¹⁷ are set according to Guidelines of Industry Classification for Listed Companies issued by the CSRC in 2001.

Model (2) tests the influence of the interaction between guarantee and high-quality external auditing on bond issuance cost. Based on prior research by Mansi et al. (2004); Zhou, Lin, Li, and Wang (2012); Lin, Li, Wang, and Liu (2013), this paper takes Credit Spread between the coupon rate of bond issuance and the treasury rate of the same period as the measure of bond issuance cost. Additionally, this paper uses Guarantees and Auditing as independent variables to test the influences of the interaction between guarantee and high-quality external auditing on bond issuance cost. Control variables such as Rating, Maturity, and Callable are selected by reference to Xu and Yang (2013). Rating is the bond rating variable, where AAA equals 4, AA⁺ equals 3, AA equals 2, and AA^- equals 1; Maturity is a debt maturity variable (three maturity dummy variables are set in the model, with four in total); Callable is a dummy variable that equals 1 if the investor has a callable option and the firm has the option of raising the coupon rate and 0 otherwise.

6. Empirical results

6.1. General characteristics of bonds over the years

Figure 1 lists the composition and development trend of the enterprise bond market in 2004–2013. In 2004, there were only four common enterprise bonds, while in 2012 there were 53 and in 2013 there were 35 until June 30. The enterprise bond, on the whole, presents a trend of rapid growth. There were 1488 enterprise bonds listed in the interbank market and stock exchange market in the period 2004–2013, including 307 common enterprise bonds, which account for 20.56%, indicating that common enterprise bonds are a low proportion of the enterprise bond market.

Figure 1. The sum of enterprise bonds issued in 2004–2013.

6.2. Sample distribution

Table 1 describes the development of bond guarantees and external auditing in common enterprise bond markets. There was only one enterprise choosing Chinese ‘Big Ten’ auditors in 2004, accounting for 25%; while in 2012 there were 20 choosing Chinese ‘Big Ten’ auditors, accounting for 40%. The results show an increase in demand for big auditors in the external auditing of the Chinese enterprise bond market. In terms of guarantee, in 2004–2007, the rate of total guarantees of enterprise bonds was over 95% due to the mandatory guarantee regulation specified by the NDRC, but the number decreased. Since 2008, the rate of total guarantees of enterprise bonds declined from 71% in 2008 to 30% in 2012. This indicates that guarantees became more flexible after the banks exited the bond guarantee market in 2008. More enterprises chose to issue bonds without guarantee, posing higher default risk in the bond market. As for bank guarantees that were the dominant guarantee type of enterprise bond prior to 2008, they were provided mainly by non-commercial banks and state-owned commercial banks. Since 2008, banks basically exited the bond guarantee market, with only a very few left to provide illegal guarantees; as for collateral, which was rare before 2008, only the enterprise bond issued by Tiger Forest & Paper Co., Ltd. on June 25, 2007 was guaranteed by such assets as land use right and real estate. With banks exiting the bond guarantee market in 2008, enterprises looked for more collateral to ensure successful issuance of bonds and reduce bond issuance cost. In terms of related-party guarantee and mutual guarantee, there were nine enterprises adopting mutual guarantee in 2008 (the year with the greatest number of cases of mutual guarantee) and eight enterprises adopting related-party guarantee in 2009 (the year with the greatest number of cases of related-party guarantee). After 2011, the quantity of related-party guarantee and mutual guarantee decreased slowly, which was possibly because enterprises were in urgent need of new guarantees to meet the issuance requirement and to reduce the costs of bond issuance after the exit of banks in 2007. With the increase of the risk awareness of the investors of related-party guarantee and mutual guarantee as well as reinforcing regulations, the use of such two kinds of guarantees has been restricted. Local-government guarantees were more frequently used in 2009 and 2010 by 9 and 11 enterprises respectively. The possible reason for this is that after cancellation of bank guarantee at the end of 2007, state-owned enterprises often used the local-government guarantee to fill the gap left by bank guarantee, with the support from local governments.

Table 1. The development of guarantees and auditing in the common enterprise bond market.

	Issued bonds	Auditing	Guarantees
			Bank Guarantee	Collateral	Related-party Guarantee	Mutual Guarantee	Local-government Guarantee	Other Guarantees	Total
2004	4	1	4	0	0	0	0	0	4
	100%	25%	100%	0	0	0	0	0	100%
2005	10	4	10	0	0	0	0	0	10
	100%	40%	100%	0	0	0	0	0	100%
2006	21	5	19	0	1	0	0	0	20
	100%	24%	90%	0	5%	0	0	0	95%
2007	32	12	30	1	0	0	0	0	31
	100%	38%	91%	3%	0%	0%	0%	0%	97%
2008	28	10	1	0	5	9	5	0	20
	100%	36%	4%	0	18%	32%	18%	0	71%
2009	44	20	1	5	8	2	9	1	26
	100%	48%	2%	11%	18%	5%	20%	2%	59%
2010	38	19	0	3	6	1	11	3	24
	100%	50%	0	8%	16%	3%	29%	8%	63%
2011	42	16	0	9	3	0	3	0	15
	100%	38%	0	21%	7%	0	7%	0	36%
2012	53	20	0	8	5	0	2	0	15
	100%	38%	0	15%	9%	0	4%	0	28%
2013	35	12	0	8	5	0	0	4	17
	100%	34%	0%	23%	14%	0%	0%	11%	49%
Total	307	120	65	34	32	12	30	9	182
	100%	39%	21%	11%	11%	4%	10%	3%	59%

Table 2 reports the sample distribution by year and industry. The annual trend in bonds suggests that the number of enterprise bonds increases throughout our sample period from 2008 to 2013. The ratio of guarantee over total bonds gradually decreased. Meanwhile, the number of mutual guarantees and local-government guarantees decreases while the number of related-party guarantee and collateral increases. From the viewpoint of industry, industry distribution is uneven: mining, manufacture and electronics/gas issue more enterprise bonds; while service and media/culture issue fewer enterprise bonds.

Table 2. Sample distribution.

Panel A: Distribution by year
	Auditing		Guarantees
	‘Big Ten’	‘Non-Big Ten’	Collateral		Related-party Guarantee		Mutual Guarantee	Local-government Guarantee		Non-Guarantee	Total
2008	10	17	0		5		9	5		8	27
2009	20	22	5		8		2	9		18	42
2010	17	18	3		6		1	11		14	35
2011	16	26	9		3		0	3		27	42
2012	20	30	8		5		0	2		37	52
2013	11	18	8		5		0	0		16	29
Total	94	133	33		32		12	30		120	227
Panel B: Distribution by industry
Agriculture				7		Information technology			2
Mining				35		Wholesale/Retailing			22
Manufacture				89		Property			5
Electronics/Gas				40		Services			1
Construction				7		Media/Culture			3
Transport/Storage				9		Multi-industry			7
Total Observations				227

6.3. Descriptive statistics

Table 3 lists the descriptive statistics of the sample. The mean of using collateral in the ‘Big Ten’ sample is 0.096, lower than 0.180 of the ‘non-Big Ten’ sample and significant at the 10% level, while the mean of the local-government guarantee in the ‘Big Ten’ sample is 0.075, lower than 0.173 of the ‘Big Ten’ sample and significant at the 5% level. This means that collateral and local-government guarantee, as positive guarantees, are significantly more likely to reduce bond credit risk and demand for high-quality external auditing. On the other hand, the mean of related-party guarantee in the ‘Big Ten’ sample is 0.192, higher than 0.105 of the ‘Non-Big Ten’ sample and significant at the 10% level. This indicates that the related-party guarantee is more likely to increase the bond default risk, which can be identified by the investors. Therefore, enterprises need external auditing to improve their reputation to ensure successful issuance and cut bond issuance cost. Furthermore, the mean of Ln(Sales) in the ‘Big Ten’ sample is 14.438, significantly higher than the 13.653 of the ‘Non-Big Ten’ sample at a level of 1%, indicating that enterprises with larger sizes tend to choose ‘Big Ten’ auditors.

Table 3. Descriptive statistics.

	Maximum	Minimum	Mean	Median	Standard deviation
Panel A: Full sample (n=227)
Auditing	1.000	0.000	0.412	0.000	0.494
Credit Spread	5.440	0.480	2.522	2.410	0.990
Related-party Guarantee	1.000	0.000	0.141	0.000	0.349
Mutual Guarantee	1.000	0.000	0.053	0.000	0.224
Local-government Guarantee	1.000	0.000	0.132	0.000	0.339
Collateral	1.000	0.000	0.145	0.000	0.353
Return-on-Assets	16.650	0.380	3.815	2.890	2.798
Leverage	0.930	0.030	0.622	0.640	0.134
Current-Ratio	5.120	0.080	1.102	1.030	0.605
Times-Interest-Earned	200.750	0.950	8.434	4.930	15.803
Ln(Sales)	18.960	9.780	13.978	13.942	1.812
Panel B: ‘Big Ten’ sample (n=94)
Credit Spread	4.790	0.560	2.296	2.195	0.944
Related-party Guarantee	1.000	0.000	0.192	0.000	0.396
Mutual Guarantee	1.000	0.000	0.053	0.000	0.226
Local-government Guarantee	1.000	0.000	0.075	0.000	0.264
Collateral	1.000	0.000	0.096	0.000	0.296
Return-on-Assets	16.650	0.590	3.795	2.870	2.856
Leverage	0.930	0.030	0.618	0.655	0.148
Current-Ratio	5.120	0.300	1.125	1.035	0.724
Times-Interest-Earned	200.750	1.780	11.405	5.715	23.328
Ln(Sales)	18.960	9.780	14.438	14.334	2.114
Panel C: ‘Non-Big Ten’ sample (n=133)
Credit Spread	5.440	0.480	2.681	2.540	0.994
Related-party Guarantee	1.000	0.000	0.105	0.000	0.308
Mutual Guarantee	1.000	0.000	0.052	0.000	0.224
Local-government Guarantee	1.000	0.000	0.173	0.000	0.380
Collateral	1.000	0.000	0.180	0.000	0.386
Return-on-Assets	12.690	0.380	3.828	2.93	2.768
Leverage	0.880	0.260	0.624	0.540	0.124
Current-Ratio	3.500	0.080	1.085	1.020	0.507
Times-Interest-Earned	37.110	0.950	6.334	4.430	5.781
Ln(Sales)	17.580	9.950	13.653	13.832	1.489
Panel D: ‘Big Ten’-’Non-Big Ten’
	Mean difference	T Value	Median difference		Z value
Credit Spread	–0.385	–2.935***	–0.345		–2.837***
Related-party Guarantee	0.087	1.768*	0.000		–1.84*
Mutual Guarantee	0.001	0.018	0.000		–0.019
Local-government Guarantee	–0.098	–2.305**	0.000		–2.153**
Collateral	–0.084	–1.870*	0.000		–1.780*
Return-on-Assets	–0.033	–0.087	–0.060		–0.103
Leverage	–0.006	–0.347	0.115		–0.184
Current-Ratio	0.040	0.490	0.015		–0.210
Times-Interest-Earned	5.070	2.063**	1.285		–2.552**
Ln(Sales)	0.785	3.098***	0.502		–2.837***

*** significant at the 1% level;

** significant at the 5% level;

* significant at the 10% level.

6.4. Pearson correlation analysis

The Pearson correlation in Table 4 indicates that with Auditing (‘Big Ten’) as the indicator of demand for high-quality external auditing, local-government guarantee and auditing have a correlation coefficient of –0.143, significant at the 5% level; collateral and auditing have a correlation coefficient of –0.118, significant at the 10% level. It indicates that local-government guarantee and collateral are more efficient in reducing demand for high-quality external auditing by the investors. On the other hand, related-party guarantee and auditing have a correlation coefficient of 0.122, significant at the 10% level, indicating that the connect-party guarantee contains higher credit risk, which increases demand for high-quality external auditing. Moreover, auditing and credit spread have a correlation coefficient of –0.192, significant at the 1% level, indicating that external auditing may reduce bond issuance cost. These results are consistent with expectations.

Table 4. Pearson correlation.

		Credit Spread	Related-party Guarantee	Mutual Guarantee	Local- government Guarantee	Collateral	State-Owned Enterprise	Return-on-Assets	Leverage	Current- Ratio	Times-Interest-Earned	Ln(Sales)	Rating	Callable
Auditing		-0.192^***	0.122^*	0.001	-0.143^**	-0.118^*	0.072	-0.014	-0.018	0.007	0.179^***	0.215^***	0.162^**	-0.041
Credit Spread			0.007	-0.138^**	-0.128^*	0.284^***	-0.487^***	0.158^**	-0.127^*	0.125^*	-0.000	-0.469^***	-0.669^***	0.201^***
Related-party Guarantee				-0.096	-0.158^**	-0.167^**	0.059	-0.034	0.189^***	-0.047	-0.112^*	-0.076	0.029	-0.247^***
Mutual Guarantee					-0.092	-0.097	0.124^*	-0.067	0.006	-0.173^***	-0.060	0.143^**	0.197^***	-0.155^**
Local- government Guarantee						-0.161^**	0.205^***	-0.129^*	-0.063	0 .009	0.079	- 0.059	0.057	-0.012
Collateral							-0.482^***	0.138^**	-0.136^**	0.055	0.015	-0.292^***	-0.186^***	0.151^**
State-Owned Enterprise								-0.383^***	0.009	-0.007	-0.002	0.215^***	0.319^***	-0.261^***
Return-on-Assets									-0.247^***	0.115^*	0.269^***	-0.144^**	-0.195^***	0.154^**
Leverage										-0.317^***	-0.453^***	0.370^***	0.161^**	-0.090
Current-Ratio											0.297^***	-0.179^***	-0.088	0.204^***
Times-Interest-Earned												0.044	0.056	0.072
ln(Sale)													0.521^***	-0.269^***
Rating														-0.313^***

*** significant at the 1% level;

** significant at the 5% level;

* significant at the 10% level.

6.5. Logistic regression of guarantees and external auditing

Table 5 shows logistic regression results of guarantees and high-quality external auditing. First, this paper categorizes guarantees provided by issuers into related-party guarantee, mutual guarantee, local-government guarantee and collateral and puts them into a logistic regression model for testing their influence on the demand for external audit. As can be seen, a related-party guarantee has a coefficient of 1.067, significantly positive at the 5% level, indicating that the investors are able to identify the credit risk of a related-party guarantee. So high-quality external auditing is required as a supplementary mechanism to disclose more information about bond risks. Therefore, firms tend to choose ‘Big Ten’ auditors. Local-government guarantee has a coefficient of –1.238, significantly negative at the 5% level, indicating that local-government guarantees (implying assurance by local government) greatly reduce investors’ expectations of bond default risk. Therefore, the investors are willing to accept ‘Non-Big Ten’ auditors with relatively lower audit quality.

Table 5. Logistic regression of guarantees on high-quality external auditing.

Variables	Estimate	Wald value
Intercept	–2.753	1.95
Related-party Guarantee	1.067	4.48**
Mutual Guarantee	–0.233	0.09
Local-government Guarantee	–1.238	4.66**
Collateral	–0.497	0.83
State-Owned Enterprise	–0.456	0.83
Return-on-Assets	–0.105	2.21
Leverage	–1.331	0.63
Current-Ratio	–0.219	0.37
Times-Interest-Earned	0.057	5.87**
Ln(Sales)	0.295	6.70**
Year	Control
Industry	Control
–2Log likelihood	253.548
	(0.001)
N	227

*** significant at the 1% level;

** significant at the 5% level;

* significant at the 10% level.

In terms of correlation between mutual guarantee and external auditing, mutual guarantee has a coefficient of –0.233 which is not significant. After further analysis of the sample, it is found that 12 bonds providing mutual guarantee are state-owned enterprises, including four issuers along with their guarantors under the sole proprietorship of central SASAC, with the remaining eight along with their guarantors controlled by local SASAC. This shows that mutual guarantee is the result of intervention and coordination of SASAC where investors can reasonably consider such guarantee to be the ‘responsibility of the government in case of bond default’. In other words, mutual guarantee here is, in fact, an implicit government guarantee, and it greatly reduces the credit risk of the mutual guarantee. Therefore, the correlation between mutual guarantee and external auditing is inconsistent with the expectation of the paper.

In terms of the correlation between collateral and external auditing, collateral has a coefficient of –0.497, indicating that there is some substitutive relation between collateral and external auditing. However, the test result is not significant. The reason may be that due to lower law enforcement efficiency in the Chinese judiciary system (Wu & Yuan, 2007), the optimal guarantee effect of collateral cannot be assured. In some cases, bond investors still need external auditing to protect their interests.

In conclusion, the above results show that a local-government guarantee is the most effective among positive guarantees, presenting a significant substitutive relationship with external auditing. A related-party guarantee has a negative guarantee effect, increasing bond default risk, thus presenting a significant supplementary relation with external auditing.

6.6. Multiple regression of guarantee, external auditing and bond issuance cost

Result (1) in Table 6 tests the influence of various guarantees and external auditing on bond issuance cost. In Result (1), the related-party guarantee has a coefficient of 0.076, indicating that the related-party guarantee will lead to an increase of credit spread, but the result is not significant. Mutual guarantee and collateral have coefficients of –0.266 and –0.129 respectively with non-significant results. Local-government guarantee has a coefficient of –0.307, significant at the 5% level, indicating that such a guarantee has the most significant effect in reducing bond issuance cost.

Table 6. Multiple regression of guarantees and external auditing on credit spread.

Variables	(1)			(2)
	Estimate	T value		Estimate	T value
Intercept	6.275	9.65***		6.191	9.54***
Related-party Guarantee	0.076	0.52		0.346	1.74*
Mutual Guarantee	–0.266	–1.19		–0.240	–0.88
Local-government Guarantee	–0.307	–2.22**		–0.323	–2.02**
Collateral	–0.129	–0.87		–0.194	–1.14
Auditing	–0.123	–1.34		–0.092	–0.77
Related-party Guarantee × Auditing				–0.485	–1.91*
Mutual Guarantee ×Auditing				–0.030	–0.08
Local-government Guarantee × Auditing				0.130	0.42
Collateral × Auditing				0.263	0.96
State-Owned Enterprise	–0.808	–5.95***		–0.810	–5.96***
Return-on-Assets	–0.033	–1.84*		–0.033	–1.86*
Leverage	–0.037	–0.08		–0.067	–0.15
Current-Ratio	0.044	0.47		0.039	0.42
Times-Interest-Earned	0.004	0.82		0.003	0.57
Ln(Sales)	–0.089	–2.67***		–0.086	–2.57**
Rating	–0.531	–8.57***		–0.529	–8.56***
Callable	–0.140	–1.09		–0.137	–1.07
Maturity	Control			Control
Year	Control			Control
Industry	Control			Control
AdjR^2	0.600			0.604
F value	14.57***			12.88***
N	227			227

*** significant at the 1% level;

** significant at the 5% level;

* significant at the 10% level.

Result (2) in Table 6 further tests the influences of interaction between various guarantees and ‘Big Ten’ on bond issuance cost. The related-party guarantee has a coefficient of 0.346, significantly positive at the 10% level, indicating that compared with non-guarantee bonds, related-party guarantee contains higher credit risk and requires higher risk premium as compensation. By contrast, the coefficient of related-party guarantee × auditing is –0.485, significant at the 10% level, indicating that compared with related-party guaranteed enterprises not hiring ‘Big Ten’ auditors, those hiring ‘Big Ten’ auditors can significantly cut their bond issuance cost. Local-government guarantee has a coefficient of –0.323, significant at the 5% level, indicating that such a guarantee can significantly cut bond issuance cost. However, the coefficient of local-government guarantee × auditing is 0.130 and insignificant, indicating that when issuers adopt a local-government guarantee, the influence of external auditing on bond issuance cost is absent. In conclusion, when issuers provide negative guarantees, external auditing serves as a supplementary mechanism to signal advantageously to the bond investors so as to significantly cut bond issuance cost; when issuers adopt positive guarantees, the effect of external auditing in cutting bond issuance cost is not significant due to the substitutive relationship between guarantee and external auditing.

6.7. Robustness test

In addition to the bond credit spread, this paper also uses the coupon rate as the measure for bond issuance cost to test the influence of the interaction of guarantees and auditing on bond issuance cost. Table 7 shows the corresponding regression results, in which the interaction coefficient of related-party guarantee × auditing is –0.676, significant at the 5% level; the interaction coefficient of local-government guarantee × auditing is 0.109 and not significant, and that of collateral × auditing is 0.454 and not significant, indicating that external auditing can significantly reduce the bond issuance interest rate for enterprises providing negative related-party guarantees but does not help much in reducing the bond issuance interest rate for enterprises adopting local-government guarantee and collateral. The above empirical results are more significant than those when using a bond credit spread as a bond issuance cost and support hypotheses 2-1 and 2-2 in this study.

Table 7. Multiple regression of guarantees, external auditing and coupon rate.

Variables	Estimate	T value
Intercept	9.870	13.68***
Related-party Guarantee	0.317	1.43
Mutual Guarantee	–0.458	–1.51
Local-government Guarantee	–0.469	–2.64***
Collateral	–0.419	–2.21**
Auditing	–0.140	–1.04
Related-party Guarantee × Auditing	–0.676	–2.39**
Mutual Guarantee × Auditing	0.429	0.99
Local-government Guarantee × Auditing	0.109	0.31
Collateral × Auditing	0.454	1.48
State-Owned Enterprise	–1.010	–6.68***
Return-on-Assets	–0.044	–2.23**
Leverage	0.218	0.44
Current-Ratio	0.062	0.60
Times-Interest-Earned	0.007	1.19
Ln(Sales)	–0.115	–3.08***
Rating	–0.466	–6.77***
Callable	–0.053	–0.37
Maturity	Control
Year	Control
Industry	Control
AdjR^2	0.571
F value	11.01***
N	227

*** significant at the 1% level;

** significant at the 5% level;

* significant at the 10% level.

7. Conclusion

Previous research shows that there are supplementary and substitutive relationships between various terms of debt contracts. When a certain term is unable to reduce the credit risk of the debtors, the creditors will require debtors to provide other terms to protect their investment interests, thus the supplementary relation between various terms is established. When a certain term is indeed able to reduce the credit risk of the debtors, the creditors lower their demand for other terms, thus establishing a substitutive relation between various terms. Following the findings of previous research, this paper empirically tests the relationship between two contract terms: guarantee and external auditing in Chinese bond market. The study results reveal that the type of guarantee has a significant relation to external auditing. Various guarantee types have distinctively different effects. A related-party guarantee is significantly positively correlated with high-quality external auditing while a local-government guarantee is significantly negatively correlated with it. For issuers providing a related-party guarantee, high-quality external auditing are significantly and negatively correlated with the bond issuance cost, while for those providing a local-government guarantee, high-quality external auditing has no obvious association with bond issuance cost. The explanation is that: although a related-party guarantee meets the guarantee requirement in form, it actually does not contribute in substance to alleviating information asymmetry. On the contrary it increases bond default risk, leading institutional investors, as major participants in the Chinese bond market, to demand external auditing more strongly to identify the credit risk. A local-government guarantee implies the credibility and assurance of local government, and hence helps to reduce the expectation of bond default by the investors, thereby reducing the demand for high-quality external auditing. Collateral has a weaker guarantee effect than that of a local-government guarantee because the protection of investors in collateral depends on the completeness of the legal system and the efficiency of law enforcement. As an emerging country in economic transition, the Chinese judicial system still has low law enforcement efficiency, and only provides limited legal protection for the investors (Wu & Yuan, 2007). This to some extent weakens the guarantee effects of collateral and forces investors to request more external auditing to reduce bond credit risk and protect their rights.¹⁸ The reason for insignificant correlation between mutual guarantee and high-quality external auditing is that a large part of issuers along with their guarantors are under the control of the central SASAC or local SASAC, implying a guarantee from the government, thus reducing the risk of such a guarantee as well as demand by the investors for high-quality external auditing. The above findings reveal that when the guarantees are negative, there is a supplementary relation between the guarantee and external auditing; when positive, they become substitutive. Moreover, the more effective the positive guarantees, the more significant the substitution between guarantee and external auditing is.

We can learn from this study that, firstly, since the Chinese bond market has not been standardized and guarantees provided by enterprises are rather in form than in substance, the regulators should pay more attention to the risks of various guarantees, especially of a related-party guarantee. Secondly, although a local-government guarantee can reduce bond default risk, its effects stem from the expectation of the investors of local governments’ insurance on bonds rather than on the credibility of the issuers themselves. It is not helpful for investors to raise their awareness for risk and may also increase the fiscal risk of local governments. Thirdly, the different influences of different guarantees on bond issuance cost indicate that investors in the Chinese bond market are capable of identifying and pricing the risks and that the bond market should form and improve its market-oriented pricing mechanism rather than just depending upon an implicit guarantee from governments. Finally, in consideration of the key role of external auditing in cutting debt agency cost, external auditing in the bond market with improving mechanisms will help to mitigate credit risk and enhance efficiency in allocation of credit resources.

Acknowledgements

We appreciate the suggestions for amendments from Professor Lijun Xia at the 2013 academic conference of China Journal of Accounting Studies. We are deeply grateful to the 2013 Annual Accounting Theory Committee of the Accounting Society of China, which awarded this paper an Excellent Paper prize. We specially thank the anonymous referees for their constructive comments. Any errors and omissions remain our own.

Notes

1. Data from China Business News on December 28, 2012. In addition, the Chinese bond market is divided into the bond market of the stock exchange and the interbank bond market. The latter is the main part of the bond market (Shi & Yang, 2011). By November 2011, the interbank bond market stock accounts for 96.2% of the total stock.

2. Although the Notice on Matters Related to Promoting Development of Enterprise Bond Market and Simplifying Procedures of Approval for Issuance promulgated by the NDRC in 2008 abandoned the regulation for the compulsory guarantee of enterprise bonds, it is specified in the Notice on Matters Concerning Further Intensifying Control over Enterprise Bonds Risk Prevention in 2012 that bond issuers applying an asset–liability ratio between 80% and 90% shall provide a guarantee in principle.

3. Although related-party guarantee and mutual guarantee have negative effects, bond regulators always use a guarantee as one of the bond regulations no matter what effects it has. For example, NDRC uses a guarantee as one of the important bond issuance conditions and the China Insurance Regulatory Commission (CIRC) requires insurance funds should be invested in bonds with a guarantee, so some enterprises still provide related-party guarantees and mutual guarantees to meet regulators’ requirements although they have no positive effects.

4. In this paper, the term ‘negative guarantees’ refers to related guarantee and mutual guarantee, which essentially do not have positive effects such as increasing bond rating and reducing interest rate, but have negative effects, e.g. increasing the default risk of issuers and their bonds. The reason that negative guarantees exist in the Chinese bond market has been explained by Note 3.

5. A third-party guarantee also includes a guarantee from credit guarantee institutions. However, currently Chinese guarantee institutions are weak and only offer a guarantee for very few enterprise bonds. The sample in this paper does not include any case of guarantees from guarantee institutions, so it falls outside the scope of the research.

6. Data from Chinabond.com.cn.

7. Smith and Warner (1979) argue that it is in the interests of the firm’s owners to include contractual provisions, which lower the costs of monitoring. They point out that observed provisions in debt contracts often include the requirement that the firm supply audited annual financial statements to the bondholders. Carey, Simnett, and Tanewski (2000) also point out that the provision of audited financial statements is normally regarded as a cost-effective contractual response to agency costs. Consequently, this paper takes external auditing as a contract term.

8. For instance, the formalities of collateral are cumbersome and the execution procedure of real security rights is complicated and time- and expense-consuming; a high audit fee is needed to hire big auditors with a high auditing quality.

9. Tang (2011) exposes how companies use auditor size as ‘true false-information’ to signal to the public to cheat outside investors. This paper has borrowed the viewpoint.

10. In 2008, China Northeast Special Steel Group Co., Ltd. had its bonds guaranteed by Benxi Steel Group Corporation under the same State-owned Assets Supervision and Administration Commission (SASAC) because it was hard for it to find a guarantor during bond issuance due to low-quality assets and weak profitability. See Northeast Special Steel Hobbled in Bond Issuance due to Leftover Problem of RMB 1.9 Billion, by 21st Century Business Herald on January 8, 2009.

11. The People's Bank of China (PBC) began complete deregulation of loan interest rate of financial institutions on July 20, 2013. For this reason the paper uses the data before June 30, 2013 for the year 2013 to prevent the impact on dependent variables of bond issuance interest and credit spread in this study.

12. Among 67 enterprise bonds from 2004–2007, 65 are guaranteed, including 63 guaranteed by banks.

13. In 227 bonds, most bonds are issued by different companies, but a few bonds are issued by the same companies. These bonds are issued in a different year, having different guarantee and financial conditions, so they could be viewed as different observations despite the same issuers.

14. Another common indicator of demand for high-quality auditing is audit fee. But issuance documents of enterprise bonds do not disclose relevant data and therefore it is impossible to test with this indicator in the paper.

15. Chow (1982) argues that potential bondholders would anticipate the wealth-transfer behaviour after the bonds have been issued and allow for these expected losses in pricing the bonds. If shareholders contract (e.g. engage external auditing) to limit their own ability to transfer wealth from bondholders, they would receive a higher price for bonds by bondholders. Chen and Li (2013) and Zhu (2013) reveal that high-quality external auditing improve the credit rating of bonds and lower bond financing cost in the Chinese bond market. Consequently, the demand of bondholders for high-quality external auditing influences bond issuing firms to choose ‘Big Ten’ even though the auditor is hired by the firm.

16. The most distinctive difference between listed companies and unlisted companies lies in corporate governance. Securities regulators impose higher requirements for corporate governance of listed companies, which significantly influences external auditing. This has been demonstrated in previous research.

17. In multiple linear regressions of guarantees, external auditing and bond issuance cost, only the 2008 annual dummy variable is set due to missing or very few interactions of guarantee and auditing for several years. Additionally, industry dummy variables with missing or few observations have been combined.

18. When the debtors have not paid off their debts, the creditors should realize their security rights on mortgaged assets via a judicial procedure.