Underwriting Syndicates and the Cost of Debt: Evidence from Chinese Corporate Bonds

ABSTRACT

This article examines the association between underwriting syndicates and the cost of debt based on a sample of Chinese corporate bonds during 2007–2013. We find strong evidence that there is a negative relationship between forming underwriting syndicates and the cost of debt. The cost of bonds is more likely to decrease when the syndicate has more members—specifically, more joint managers. Additionally, by measuring the information asymmetry using several methods, we observe that this negative relationship is more pronounced when the information asymmetry between issuers and bond investors is more serious. The above results are robust after controlling for the potential endogeneity by constructing instrumental variables based on the unique setting of China’s corporate bond market.

KEY WORDS:

• Cost of debt
• information asymmetry
• underwriting syndicates

Acknowledgments

We greatly appreciate the two anonymous referees’ valuable comments. We thank Prof. Sangho Yi from Sogang University for his important advice on the improvement of this paper. We also want to show our gratitude to Prof. Guangming Gong and Dr. Yong Ma from Hunan University for their useful suggestions. Moreover, discussants and all audiences of this paper in the 2015 Conference on Cross-strait Banking and Finance also contributed to this paper. All errors and omissions are the sole responsibility of the authors.

Funding

Shenggang Yang acknowledges the funding support from the innovation research group “Financial Engineering and Risk Management”, founded by the National Natural Science Foundation of China [grant number 71221001]; Xun Gong and Si Xu acknowledge the funding support from the Ph.D. Program Foundation of Ministry of Education of China [grant number 20130161110045].

Notes

^1. One of the most obvious differences is that the creditors of syndicated loans have to directly bear the default loss, but the members of underwriting syndicates are mostly concerned with their reputation.

^2. In China, the CSRC requires that all members of underwriting syndicates sign a Joint Underwriting Agreement before working together, which states each member’s responsibility, allocation of underwriting shares and issue fee, etc. However, this agreement is not required to be publicly disclosed. We have tried to collect data by communicating with some investment bankers, but most refused to share this information. They also stated that this information would be disclosed only when they believe it may attract more potential investors or benefit the enquiry procedure.

^3. The Wind and CSMAR databases were developed according to the international standards of databases to meet the requirement of academic research, and their databases were used for several recent studies. See, for example, Chen et al. (2010) and Chen and Zhu (2013).

^4. The Securities Association of China publishes a report every year that ranks securities companies based on different indicators, such as size, equity underwritten volume or market value, and corporate bond underwritten volume or market value. In this article, we choose the ranking based on the corporate bond underwritten volume and use the ranking based on the corporate bond underwritten market value for a robustness check.

^5. For corporate bond-issuing procedures in China, the underwriting syndicate members include the book manager, joint managers, co-managers, and retailers. Because the retailers do not participate in the design of debt contracts, we do not consider them to be substantial members of the underwriting syndicates in this article.

^6. For brevity, we do not present the details of the calculations. Every step follows the instructions of Dechow and Dichev (2002), except we use the negative value of this variable for ease of interpretation.

^7. In our sample, all cross-listed companies are cross-listed in both the Chinese mainland and Hong Kong. No bond issuer is cross-listed in other markets, such as the USA or Europe.

^8. In the developed markets, bonds can be issued at a discount or at a premium. Therefore, some existing studies use two variables, coupon spread and issuing price, to jointly measure the cost of corporate bonds (Livingston and Miller 2000). However, all Chinese corporate bonds are required to be issued at their par value by the CSRC, which means that discount or premium issuing is forbidden. Hence, we believe that the coupon rate can accurately measure the real financing cost of the bond. We thank the referee who suggested us to discuss this issue in detail.

^9. Many issuers run seasoned equity offerings several times. The date used here is that of the issuers’ nearest SEO before the corporate bond issuance. This variable is included to control for the spillover effect of the nearest equity issuing.

^10. We thank the referee for the valuable suggestion.

^11. The bond ratings shown in Table 1 are not issued by one rating agency. Almost all credit rating reports of corporate bonds are issued by the five big domestic rating agencies, which are CCXI Credit Rating, China Lianhe Credit Rating, Shanghai Brilliance Credit Rating, Dagong Global Credit Rating, and Pengyuan Credit Rating. In untabulated robustness tests, we further control for the influence of different rating agencies. The results still support all our conclusions. We thank the referee for the valuable suggestion.

^12. The average number of co-managers in the syndicates in the US equity market is 1.84 (Popescu and Xu 2011) from 1993 to 2005 and 2.44 (Jeon and Ligon 2011) from 1997 to 2007.

^13. Duration is defined as the natural logarithm of the number of months between the date of the issuer’s equity IPO and the date bond issued. Therefore, the average number of months between the firm IPO date and bond issuing date equals e^4.370= 79.04, which is about 6.6 years.

^14. As the setting we investigate is underwriting syndicates, only the book manager, joint managers, and co-managers with whom the issuers previously cooperated for their equity offering are considered previous underwriting relationships.

^15. Corwin and Schultz (2005) indicate that syndicate members may compete with book managers to obtain future business with the issuer.

^16. As mentioned before, no more than 10 default cases have existed until now, and none were included in our data. Hence, we predict that the potential loss of reputation will not be an important concern for book managers.

^17. Generally, book managers decide whether to form syndicates and who the potential partners will be at a very early stage of the underwriting procedure. By reading Board Resolution on Bond Issuing and Resolution of Shareholders Meeting on Bond Issuing, we find that the bond is underwritten by a sole underwriter or a syndicate. However, the formal rating report is published at much later stage, generally when the book manager creates an issuance prospectus. Therefore, we believe that the book manager cannot access a formal credit rating report as a reference to support the decision of forming a syndicate.

^18. We thank the referee for the comments on this issue, which advised us to further discuss this issue in detail and check the robustness of our IV regression results.

^19. We include the firm-level and bond-level control variables in the selection model for the basic PSM method, and furthermore, we include Uw_Channel and Uw_Compet in the selection model for the extended PSM method.

^20. The other methods (the NEAREST NEIGHBOR and the KERNEL matching approach) present similar results that show that the Treated Group and Control Group are well balanced. For brevity, we do not report these results, but they are available upon request from the authors.

^21. One limitation of the PSM technique is that the selection variable should be a dummy variable, which cannot divide the original sample into two subsamples. Because Uwsyn_Num is a count variable, we cannot use the PSM technique to account for its potential endogeneity issue.

^22. Similar to the treatment in the previous section, we also run the IV regression to control for the potential endogeneity using the instrumental variables, and the results are still qualitatively similar.

^23. This is an equity index, similar to a Chinese version of the S&P 500. Firms in this index are well represented the market because they constitute almost 60% of the market value in the Chinese A-share market.

^24. We thank the referee for the valuable suggestion.

^25. We thank the referee for the valuable suggestion.

Additional information

Funding

Shenggang Yang acknowledges the funding support from the innovation research group “Financial Engineering and Risk Management”, founded by the National Natural Science Foundation of China [grant number 71221001]; Xun Gong and Si Xu acknowledge the funding support from the Ph.D. Program Foundation of Ministry of Education of China [grant number 20130161110045].