Abstract
Although China’s bond market is expanding quickly and on a bigger scale than the stock market, it still faces a number of issues. One issue that restricts its long-term development is the relatively small scale of credit bonds. For the growth of credit bonds, it is necessary to boost the credit rating market first. In this study, it is assumed that the higher the credit rating of the bonds, the lower the issue cost will be. To prove this hypothesis, an empirical analysis has been made herein on the influence of credit ratings on the bond floatation market.
1. A Design of Model
1.1. An Overview of the Model
The growth of the credit rating market in China cannot be separated from that of the bond market. The issuance of treasury bonds was resumed by the Ministry of Finance in the 1980s. Since then, the bond market has thrived for nearly 40 years. Compared with stocks, bonds have evident financing advantages, such as lower costs due to the early interest payments, no dilution of shareholders’ equity, and no change of control [1, 2]. In addition, the issuance requirements of bonds are easier to meet than those of stocks, making financing more convenient.
Credit ratings have an impact on the bond market, which is mainly reflected by the relationship between the bond issue cost and its credit rating [3–5]. Investors who are unfamiliar with the issuers may find it challenging to gain a thorough knowledge of the issuers’ conditions and credit risks. Credit rating agencies use a variety of public and nonpublic information to rate issuers or projects [6–9]. Their credit ratings provide investors with a channel to know the issuers. Before making decisions about their investments, investors will read authentic credit ratings. If the rating results have a sufficient impact on the bond market, investors will prefer bonds with high credit ratings. Therefore, bonds with high ratings are more popular among buyers. The higher a bond’s rating, the lower the issuance cost it will carry.
Western scholars began their research on credit ratings as early as in the 1960s. They have conducted in-depth discussions on the default rate, yield, rating methods, and many other fields. As a result, they have gathered numerous valuable findings that have been applied to the rating market. Foreign literature on credit rating can be divided into four categories: the relationship between credit rating and default rate; the influence of credit rating on bond yield; the distinctions between rating agencies; and the rating methods. Only in recent years have Chinese scholars begun to pay attention to credit ratings, most of which are comprehensive studies and institutional comparative studies, and few are quantitative studies. China’s credit rating research can be divided into two categories: comprehensive studies on the credit rating market in China and other countries, and through comparison, suggestions to China’s credit rating market, and research on rating methods. Quantitative studies on credit ratings are rarely conducted by Chinese academics.
At present, many foreign literature have adopted the econometric regression method to analyze the issuance cost and influencing factors of bonds [10–14]. The most popular models include NIC (net interest cost) model, TIC (true interest cost) model, and IFR (internal financing rate) model. Though they are roughly the same in economic significance, the models are different in research purposes and backgrounds. The NIC model is often used to analyze bonds that are issued in the same way with a focus on the net interest cost. The explanatory variables include scale, issuance cost, call option, market interest rate, market volatility, credit rating, regional factors, and competitive scalar. The TIC model and IFR model neglect the changes in the economic cycle to use the real interest cost, which is more applicable than the net interest cost. TIC is measured by the real interest cost. The explanatory variables include the issuance volume, issuance period, call option, whether it meets the bank’s preferential conditions, effective state government tax difference, bond index, index volatility, issuance frequency, issuance purpose, pricing method, and credit rating. The difference between the internal financing interest rate and the real interest cost is used by the IFR model as the explanatory variable. The explanatory variables use the bond type, pricing method, credit rating, bond insurance, whether it is a repayment bond, issuance period, and issuance scale.
As China has a short research history in the bond market, few studies have used the econometric model to analyze the relationship between the credit rating and the bond market [15–18]. He and Jin [19] used the “True Interest Cost” model and the Internal Financing Rate to measure the issue cost of the bonds. The calculation formula is as follows: where BP = the actual amount of money raised by the issuer, i.e., the nominal value less discount or plus premium; Ci = the interest to be paid on the bonds during the period i; Fi = the principal amount of the bonds payable during period i; TIC = the true interest cost.
This empirical study has selected the credit rating, bond issuance scale, bond maturity, bond market index, market volatility, the issuer’s call option, the investor’s put option, the place of issuance, and the industry of the issuer as independent variables to construct the following linear regression model: TIC = α (rating dummy) + β (other arguments) + error term.
This model compares the coefficient symbol of the rating variable and its absolute value. If the absolute value of the high-rating variable coefficient is greater than the low-rating ones, it indicates that credit rating will affect the cost of issuing bonds and proves that credit rating influences the China’s bond market.
1.2. Model Selection and Design
Based on the actual economic situation of China, this paper adopts a modified version of the TIC model and selects the difference between yield to maturity and benchmark interest rate on the closing date of the bond year as the credit spread and dependent variable. In this way, it not only includes the economic samples with a long economic cycle but also reflects various types of bond payments. Moreover, it establishes a multiple linear regression model about bond issuance spread, to study the influence of credit rating factors on bond issuance spread.
This paper selects the issue spread (TIC) as the explanatory variable, and credit rating, short-term credit rating, and long-term (subject) credit rating as the explanatory variables (their ratings were linearly converted). Control variables include the bond year, bond type, actual issuance, issue period, callable option, puttable option, interchangeability, cross-market, annual amplitude, and cumulative trading days of this year. The adopted linear regression equation is as follows:where = the residual error; = the regression coefficient; TIC = the issue spread, which is the difference between the yield to maturity and the benchmark interest rate on the closing date of the bond year; Ltc r = the credit rating, which refers to the rating of the bond (debt rating); Trdyer = the rating of the bond issuer (entity rating).
Table 1 shows the detailed information about the variables.
Based on the rating classification of S&P, Fitch, and Moody’s, this paper processes the data of bond ratings and linearly converts the rating data into continuous variables.
1.3. Selection and Source of the Data
In this paper, a total of 97,690 bonds issued in CSMAR from 2009 to 2019 are selected as the population sample, including 12,700 bonds in Shanghai Stock Exchange, 5,447 bonds in Shenzhen Stock Exchange, and 79,543 bonds in the interbank market. Among them, 40,367 bonds have not yet expired, and 57,323 bonds have expired. The paper only retains enterprise bonds and corporate bonds and excludes other types of bonds in China’s bond market, such as the national bonds or local government bonds. After matching the annual trading information of bonds and their ratings, 3,166 bonds are finally retained, among which 2,783 are corporate bonds, 383 are enterprise bonds, 2,821 bonds are in Shanghai Stock Exchange, and 345 bonds are in Shenzhen Stock Exchange, and 2,963 bonds have not yet expired, and 203 bonds have expired.
1.4. Study Methodology
This paper is divided into five parts sections:
The first is the research methods and data, which includes theoretical research, hypothesis, model setting, variable interpretation, and data sources. Second is the statistics description. The third is the measurement of the credit spread. Forth is the analysis of the results and summary. Fifth is the analysis of the limitations and suggestions based on the research results.
2. Descriptive Statistics
According to the descriptive statistics in Table 2, the average value of credit spread is 16.91, the standard deviation is 571.81, the minimum value is -67.44, and the maximum value is 42,114.54. This indicates that the credit spread is very uneven. Therefore, this paper winsorizes the credit spread, with the critical percentage of 0.05. After winsorization, the standard deviation uniformity of the credit spread is greatly improved, with the mean value, standard deviation, minimum value, and maximum value of the credit spread to be 6.32, 3.22, 3.06, and 16.33, respectively. The average value of credit rating, short-term credit rating, and long-term credit rating is higher than 15, indicating that Chinese bond ratings are at a high level.
3. Discrepancy Analysis of Credit Spread
3.1. Credit Spreads of Different Bond Types
According to the descriptive statistics in Table 3, the independent sample T test is adopted to measure the credit spread of different types of bonds. The sample size of the credit spread of corporate bonds is 5,629, with an average value of 6.241 and a standard deviation of 0.041. The sample size of the credit spread of enterprise bonds is 7,300, with an average value of 6.916 and a standard deviation of 0.146. The T value obtained by the independent sample T test is −5.336, and the corresponding value is 0.000. |T| >= 2.58 indicates that it is significant at the significance level of 0.01. When value is less than 0.01, it means that there is a very significant statistical difference in the credit spread of different types of bonds at the significance level of 1%. In other words, the credit spread of enterprise bonds is significantly higher than that of corporate bonds.
According to Figure 1, the credit spread of enterprise bonds is significantly higher than that of corporate bonds. This is because enterprise bonds are issued by institutions affiliated with central government departments or by enterprises with state-owned assets. Therefore, credit ratings exert a great influence on investment decisions on such bonds. Moreover, investors prefer the rating of bond issuers when referring to the credit ratings of corporate bonds. Corporate bonds with higher credit ratings require significantly lower issuance cost.
3.2. Is the Credit Spread Callable?
The independent sample T test is adopted to measure the credit spread of callable bonds and noncallable bonds. According to the statistical results (Table 4), the value is 0.000, indicating a very significant statistical difference in the credit spread of callable bonds and noncallable bonds. |T| >= 2.58 means that it is significant at the significance level of 0.01. In other words, the credit spread of noncallable bonds is significantly larger than that of callable ones.
As Figure 2 shows, issuance clauses of bonds will affect the credit spread. Generally, noncallable bonds are more stable for bond investors, because when compared with callable bonds, they generate a steadier income. Credit ratings exert more influence on noncallable bonds’ credit spreads but serve as only one of the influencing factors to callable bonds. This is because callable bonds are affected by various clauses or special clauses when being issued. Therefore, credit ratings will have a less significant influence on callable bonds than the noncallable ones. Noncallable bonds, which have higher credit ratings, require lower issue costs.
3.3. The Credit Spreads of Puttable and Nonputtable Bonds
The independent sample T test measures the credit spreads of puttable bonds and nonputtable bonds. According to Table 5, the value is 0.000, indicating a very significant statistical difference in the credit spreads of puttable bonds and nonputtable bonds. |T| >= 2.58 means that it is significant at the level of 0.01. In other words, the credit spread of nonputtable bonds is significantly higher than that of puttable ones.
Similarly, when puttable bonds are issued, they are added with special clauses. As a result, credit ratings are no longer the only influencing factor in this kind of bond. Due to the presence of additional puttable options, credit ratings’ influence is not as significant as they are to the nonputtable bonds. For investors, the yield of a nonputtable bond is more stable. Therefore, credit ratings have more significant impact on nonputtable bonds’ credit spreads as shown in Figure 3.
Thus, when bonds are issued, credit ratings will affect the credit spreads of bonds with special clauses, but their influence is much higher than bonds without special clauses.
3.4. The Credit Spreads of Cross-Market Bond
The independent sample T test is adopted to measure the credit spreads of cross-market bonds and non-cross-market bonds. According to Table 6, the value is 0.000 indicating a very significant statistical difference in the credit spreads of cross-market bonds and non-cross-market bonds. |T| >= 2.58 indicates that it is significant at the level of 0.01. The credit spread of non-cross-market bonds is much higher than that of cross-market bonds.
The bond markets can be categorized into the interbank bond market and the exchange bond market, among which the interbank market is more frequently used for bond issuance and trading. According to Figure 4, the more liquid interbank bond market requires lower transaction fees for a larger trading scale while the exchange market charges a higher issuance cost. Cross-market bonds, such as treasury bonds and enterprise bonds have high ratings and can be traded across the two markets while other types of bonds cannot. Therefore, the non-cross-market bonds have a significantly higher credit spread than the cross-market bonds.
According to the results in Table 7, when other conditions remain unchanged, there is a very significant negative relationship between credit spread and credit rating, a significant negative relationship between credit spread and bond year, no significant relationship between credit spread and bond type, a significant negative relationship between credit spread and actual issuance, a significant negative relationship between credit spread and duration, a significant positive relationship between credit spread and call option, a significant negative relationship between credit spread and put option, a significant positive relationship between the credit spread and cross-market, no significant relationship between credit spread and annual amplitude, and a significant positive relationship between credit spread and accumulated trading days of this year.
The results of model (2) ∼ (4) are consistent with the result of model (1), indicating a significant, robust, and negative relationship between credit spread and credit rating.
The independent sample T test shows that the credit spread of enterprise bonds is significantly higher than that of corporate bonds; the credit spread of noncallable bonds is significantly higher than that of callable bonds; the credit spread of nonputtable bonds is significantly higher than that of puttable bonds; and the credit spread of non-cross-market bonds is significantly higher than that of cross-market bonds. The regression analysis from model (1) ∼ (4) shows a significant and negative relationship between credit spread and credit rating, short-term credit rating and long-term (subject) credit rating, when other conditions remain unchanged.
Therefore, it is concluded that the higher the issuer’s credit rating, the lower the credit spread will be. Credit rating, which can be divided into issuer rating and debt rating, has been recognized by the Chinese bond market as a method to evaluate credit risk. Both types of credit ratings have an impact on the issue cost, but the impact of debt rating is stronger. This indicates that investors are more concerned about the risk of the bond itself during bond investment. The acceptance of credit ratings among investors also shows that the credit rating market plays an important role in the development of the bond market by reducing information asymmetry and offering a reliable basis for investors to make decisions.
4. Conclusion
Based on the TIC model, the impact of credit ratings on the bond market has been examined in this research. It begins by adopting a modified TIC model based on the current circumstances in China. Second, it selects the data source and describes the statistics. Third, it performs a discrepancy analysis of credit spread and obtains the findings. This paper has some limitations as well. For example, the credit rating is relatively centralized, and rating agencies need more credibility. These are also the key directions for future research.
Data Availability
The data used to support the findings of this study are within the article.
Conflicts of Interest
The authors declare that there are no conflicts of interest.