Loan guarantees and the cost of debt: evidence from China

ABSTRACT

In this article, we examine the potential influence of loan guarantees and the nature of ownership on a company’s cost of debt. Using data on Chinese A-share listed companies from 2007 to 2014, we find that guaranteeing another entity’s debt significantly increases the guarantor’s cost of its own debt. Regarding the nature of ownership, our results indicate that the cost of debt for state-owned enterprises (SOEs) is lower than that for non-SOEs. Among SOEs, firms controlled by the central government have lower cost of debt than firms controlled by local governments. We also find some evidence that local government ownership mitigates the effects of loan guarantees on the cost of a guarantor’s own debt.

KEYWORDS:

• Loan guarantee
• cost of debt
• state-owned enterprises
• information asymmetry

JEL CLASSIFICATION:

• E43
• G21
• G28
• G32

Acknowledgements

We gratefully acknowledge helpful comments received form participants at the 2014 American Accounting Association annual meeting in Atlanta.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ The data on capital financing are from the China Statistical Yearbook of 2014.

² 2013 is not an unusual year. Over the period from 2002 to 2013, capital raised from banks, bond issuances and equity offerings represented 85.5%, 8.9% and 2.9% of total capital raised, respectively.

³ Figures do not add to 100% because there is a small ‘other’ source of capital for Chinese companies.

⁴ Guarantors need to bear joint and several liabilities for their credit guarantees in China.

⁵ Beck, Klapper, and Mendoza (2010, 10) note that ‘it is often difficult for banks to conduct risk assessments, since data might be sparse and of limited reliability as [the primary debtor’s] financial statements are generally not audited’.

⁶ Banks may also be less diligent in their monitoring of the use of the loan proceeds by the primary debtor when the loan is guaranteed. Such reduced monitoring could benefit the primary debtor because the borrower could use the loan proceeds for items not anticipated by the loan agreement, such as perquisite consumption or projects with excessively high risk. Such actions could benefit the primary debtor, but would have an incremental adverse effect on the guarantor company, who would then have an incentive to more closely monitor the use of the loan proceeds. In effect, loan guarantees could serve to transfer a portion of the bank’s monitoring responsibility to the guarantor company.

⁷ Behaviour in which benefits accruing to a controlling shareholder, while imposing harm on the company and its other shareholders, is often referred to as ‘tunnelling’.

⁸ The Basel III accords are designed to provide guidance to banking regulators on measuring capital sufficiency in banks.

⁹ The full name for CSMAR database is China Stock Market & Accounting Research database. It is the largest economic and financial research database in China and it is the only database included by Wharton Research Data Services (WRDS) in greater China region.

¹⁰ Wind database provides real-time fundamental data, exchange data, earnings estimate data and market data for financial professionals. It serves more than 90% of financial institutions in China.

¹¹ In untabulated analyses, we also winsorized at the 3rd and 97th percentiles and at the 5th and 95th percentiles. Our results were not materially affected by these different levels of winsorization.

¹² Data may be available for years before the company was listed due to disclosure requirements surrounding a company’s initial public offering (IPO).

¹³ Consistent with Kim et al. (2011), tangible assets are computed as the difference between total non-current assets and intangible assets.

¹⁴ The classification of industry is based on industry classification of China Securities Regulatory Commission (CSRC). Because of the large number of observations in manufacturing, we sub-classify these firms by their secondary industry.

¹⁵ The largest correlation coefficient in Table 3 is between SOE and LOCAL is (0.649). However, these two variables are not in the same regression model, which can reduce collinearity concerns.

¹⁶ We computed variance inflation factors (VIFs) in all of our models. The largest variance inflation factor (VIF) is 1.99, suggesting that multicollinearity is not a serious problem in our models.

¹⁷ The standard deviation of the coefficient on GUA is 0.036, and the mean of COSTDEBT is 2.272. Following the calculation of Ball, Jayaraman, and Shivakumar (2012), the economic significance will be 0.746% (i.e. 0.471*0.036/2.272).

¹⁸ The calculations are similar to that for GUA: (−0.330*0.048/2.272 = −0.697%, −0.153*0.041/2.272 = −0.276%).

¹⁹ The excluded base category is non-SOE*GUA. The coefficients on the interaction variables therefore measure the influence of central and local SOEs relative to non-SOEs.

²⁰ We use 1-year deposit rate of central bank of China as the benchmark interest rate. In fact, when we use 1-year loan interest rate of central bank of China, the results still remain.

Additional information

Funding

This work was supported by the National Natural Science Foundation of China [Projects number: 71172186, 71472148, 71102095].