Optimistic credit rating and its influence on corporate decisions: evidence from Korea

* Accepted by Hong Hwang upon recommendation by Feng Gu.

Abstract

Credit rating agencies have often been a target for criticism due to their inaccuracy and untimeliness. This paper sheds light on this issue by examining how an optimistic credit rating influences corporate decisions. We use Korean corporate credit ratings and construct the measure of credit rating optimism as the deviation of the actual ratings from benchmark ratings based on US corporate ratings. We find that rating optimism has a negative association with cost of debt and a positive relation with debt financing and investment. We also find that a positive relation between investment and future performance is weaker for firms with optimistic ratings. This finding suggests that inaccurate credit ratings would damage efficient capital allocation in the capital market.

Keywords:

• Credit rating
• rating optimism
• cost of debt
• external financing
• investment

View correction statement:

Corrigendum

Acknowledgement

We gratefully appreciate helpful comments from Hong Hwang (Editor), and the anonymous reviewer. This paper is based on the dissertation of Seungbin Oh. Seungbin Oh is grateful for the advice and comments of his dissertation committee: Lee-Seok Hwang (Advisor), Bok Baik (Chair), Moonchul Kim, Woo-jong Lee, and Young-Jun Kim.

Disclosure statement

No potential conflict of interest was reported by the authors.

Correction Statement

This article was originally published with errors. This version has been corrected. Please see Corrigendum (https://doi.org/10.1080/16081625.2018.1432467).

Notes

1. Majority shares of two CRAs in Korea (Korea Ratings and Korea Investors Service) are held by global CRAs (Fitch and Moody’s, respectively). Another CRA in Korea, Nice Rating, has a strategic relationship with R&I, a Japanese CRA.

2. Even in US debt market, there is a concern that CRAs make biased decision on credit rating due to conflict of interests (e.g. Liao, Li, and Hung 2017).

3. In this paper, we use ‘inflation of credit ratings,’ ‘rating inflation,’ or ‘rating optimism’ interchangeably.

4. Our result using bond spread is different from the findings based on US companies that debt investors can partially undo the impact of rating conservatism (Baghai, Servaes, and Tamayo 2014). This inconsistency suggests a possibility that, relative to the US debt market, the Korean debt market is likely to be less efficient.

5. Bank of Korea (2013) presents that Korean credit ratings are more generous than those in the US. Below table compares credit ratings issued by both the US and Korean credit rating agencies for the same major Korean companies and presents that the credit ratings issued by Korean agencies are actually higher than those issued by US agencies.

Source: The bank of Korea, 2013, Financial stability report.

6. The halo effect of the Korean business group in the debt market has long been criticized for potentially affecting the credit rating inflation of chaebol firms. In 2014, a controversy over the practice of incorporating the business group effect in affiliates’ credit rating was provoked by the default event of KT-ENS. KT, the parent firm of KT-ENS and one of the major telecommunication companies in Korea, made no financial support to KT-ENS. To mitigate the halo effect of business group in the credit rating, the Korean financial regulatory body (i.e. Financial Services Commission) mandated that the Korean credit rating agencies issue stand-alone credit ratings of subsidiaries from 2018.

7. There is a possibility that interwoven relations among affiliates within a business group may work as a channel through which one affiliate’s risk is quickly transferred to the other affiliates. In this case, all firms within the same business group are likely to be downgraded, suffering from risk contagion rather than enjoying co-insurance benefits. Several anecdotal evidence (i.e. Tongyang, STX, and Woongjin) shows that one affiliate’s default triggered simultaneous plunges in credit ratings of all the other affiliates. It is also likely that the close business ties make it difficult for one good performing affiliate to get a higher credit rating, completely isolated from the other affiliates.

8. Issuer-level credit ratings (hereafter credit ratings) represent CRAs’ current opinion of an obligor’s overall financial capacity to pay off its financial obligations (S&P 2009). On the other hand, issue-level credit ratings reflect only positive or negative adjustments relative to the issuer’s ratings, depending on the factors related to specific financial obligations, i.e. maturity of underlying instruments, priority rankings and agency problems between debtors and creditors or between debt holders and shareholders (S&P 2009).

9. COMPUSTAT (DATAGUIDE) provides credit ratings data from 1985 (1997).

10. Detailed definitions of the variables are given in Appendix 1.

11. Several prior studies suggest that macroeconomic conditions have considerable impacts on the default risk (Hackbarth, Miao, and Morellec 2006). They argue that an economic expansion (e.g. increase in GDP growth rate) increases a firm-level growth rate and hence decreases the default probability. Since the default probability influences the credit ratings, it is highly probable that the macroeconomic conditions would affect the credit ratings. Although the year and industry fixed effects in Equation (1(1) $$Z_{i,t}={\mathit{\alpha }}_t+{\mathit{\beta }}^{\prime }{X}_{i,t}+{\mathit{\epsilon }}_{i,t}$$(1) ) indirectly control for the macroeconomic factors, we include GDP growth rate and per capita GDP growth rate instead of year fixed effects in Equation (1(1) $$Z_{i,t}={\mathit{\alpha }}_t+{\mathit{\beta }}^{\prime }{X}_{i,t}+{\mathit{\epsilon }}_{i,t}$$(1) ) and re-estimate the benchmark credit ratings and rating difference. The untabulated results using the re-estimated rating differences with the macroeconomic factors do not alter the findings to be presented in the rest of the paper.

12. The results are robust to different model specifications such as 5-year rolling, yearly, and pooled regressions.

13. Baghai, Servaes, and Tamayo (2014) argue that including the actual ratings addresses the concern that rating conservatism could proxy for an omitted variable in the rating model because the firms’ actual credit rating reflects all the information employed by the CRAs. However, we add the expected credit ratings instead of the actual credit ratings due to a high correlation between actual credit ratings and the difference variable (0.524), which raises a concern of multicollinearity. However, the overall inference remains unchanged when we use the actual credit rating as a control variable.

14. The descriptive statistics for key variables are presented in Appendix 2 for the sake of brevity.

15. This result might also imply that the Korean debt market consists of firms with lower credit risk than the US debt market. However, our results of the real impact of rating optimism on corporate decisions might reduce such concern (see Sections 4.2 and 4.3 of the paper).

16. We use the yearly difference of sovereign risk between the US and Korea and the annual averages of exchange rate between the US dollar and Korean won as control variables.

17. To examine the effect of business group affiliation on CRAs’ rating process, we choose two variables: N_FIRM represents the number of member firms within a business group, and RAL_AT, measured by the sum of total assets of other affiliated firms deflated by the focal firm’s total assets, refers to the relative size of resources a firm could utilize from its own business group.

18. The similar explanatory powers in the expected credit rating models using the US and Korea data suggest that the impacts of unknown omitted variables on US and Korean credit ratings are also likely to be comparable.

19. The results remain unchanged when we transform a raw value of Dif into a quartile variable to mitigate the impact of nonlinearity and outliers.

20. To gain further insight on whether the bond market participants understand the optimistic credit rating information and correct their decisions in the future, we additionally test whether the bond yield spreads in t + 1, t + 2, and t + 3 continue to be negatively associated with the rating optimism in t − 1. Untabulated results discover that the negative impact of rating difference in t − 1 on bond spreads is statistically significant for t, t + 1, and t + 2 but not for t + 3. Moreover, the magnitudes of coefficients decrease as the rating optimism becomes outdated. Combined, these results suggest that bond market participants do not see through the credit rating optimism in the near future.

21. We calculate the economic significance of rating optimism as follows: 0.03 (coefficient on Dif in column (2)) × 0.046 (mean of INVEST) × 298,000 million won (mean of total asset) = 411 million won.

22. To control for different firm characteristics that may affect a firm’s investment decisions, we select matched sample firms having similar probability of over-investment. In constructing the matched sample, we incorporate the variables which affect the likelihood of over-investment based on Biddle, Hilary, and Verdi (2009). And the final regression model in Table 6 is based on Beatty, Liao, and Yu (2013).