Abstract
In this paper, using the measures of the credit risk price spread (CRiPS) and the standardized credit risk price spread (S-CRiPS) proposed in Kariya’s (A CB (corporate bond) pricing model for deriving default probabilities and recovery rates. Eaton, IMS Collection Series: Festschrift for Professor Morris L., 2013) corporate bond model, we make a comprehensive empirical credit risk analysis on individual corporate bonds (CBs) in the US energy sector, where cross-sectional CB and government bond price data is used with bond attributes. Applying the principal component analysis method to the S-CRiPSs, we also categorize individual CBs into three different groups and study on their characteristics of S-CRiPS fluctuations of each group in association with bond attributes. Secondly, using the market credit rating scheme proposed by Kariya et al. (2014), we make credit-homogeneous groups of CBs and show that our rating scheme is empirically very timely and useful. Thirdly, we derive term structures of default probabilities for each homogeneous group, which reflect the investors’ views and perspectives on the future default probabilities or likelihoods implicitly implied by the CB prices for each credit-homogeneous group. Throughout this paper it is observed that our credit risk models and the associated measures for individual CBs work effectively and can timely provide the market credit information evaluated by investors.
Similar content being viewed by others
References
Bharath, S. T., & Shumway, T. (2008). Forecasting default with the merton distance to default model; Forecasting default with the Merton distance to default model. Review of Financial Studies, 21, 1339–1369.
Chava, S., & Jarrow, R. A. (2004). Bankruptcy prediction with industry effects. Review of Finance, 8, 537–569.
Duan, J. C., Sun, J., & Wang, T. (2011). Multiperiod corporate default prediction—a forward intensity approach. RMI working paper, No. 10/07, National University of Singapore.
Duffie, D. (2011). Measuring corporate default risk. Clarendon lectures in finance. Oxford: Oxford University Press.
Duffie, D., & Singleton, K. J. (2003). Credit risk pricing, measurement, and management. Princeton: Princeton University Press.
Hillegeist, S. A., Keating, E. K., Cram, D. P., & Lundstedt, K. G. (2004). Assessing the probability of bankruptcy. Review of Accounting Studies, 9, 5–34.
Kariya, T. (1993). Quantitative methods for portfolio analysis. Berlin: Springer.
Kariya, T. (2013). A CB (corporate bond) pricing model for deriving default probabilities and recovery rates. Eaton, IMS Collection Series: Festschrift for Professor Morris L.
Kariya, T., & Kurata, H. (2004). Generalized least squares. New York: Wiley.
Kariya, T., Wang, J., Wang, Z., Doi, E., & Yamamura, Y. (2012). Empirically effective bond pricing model and analysis on term structures of implied interest rates in financial crisis. Asia-Pacific Financial Markets, 19, 259–292.
Kariya, T., Yamamura, Y., Inui, K., Wang, Z., & Takada, H. (2014). Measuring credit risk of individual corporate bonds and deriving term structures of default probabilities (submitted).
Kariya, T., Yamamura, Y., & Wang, Z. (2016). Empirically effective bond pricing model for USGBs and analysis on term structures of implied interest rates in financial crisis. Communications in Statistics-Theory and Methods, 45(6), 1580–1606.
Koopman, S. J., Lucas, A., & Monteiro, A. (2008). The multi-state latent factor intensity model for credit rating transitions. Journal of Econometrics, 142, 399–424.
Lando, D. (2004). Credit risk modeling. Princeton: Princeton University Press.
Saunders, A., & Allen, L. (2002). Credit risk measurement. New York: Wiley.
Shumway, T. (2001). Forecasting bankruptcy more accurately: A simple hazard model. Journal of Business, 74, 101–124.
Acknowledgments
This research was supported by the Japan Society for the Promotion of Science (JSPS), Grant-in-Aid for Scientific Research (A), No. 23243040. The authors also thank Professors T. Lai of Stanford University and N. O’Reilly of Rutgers University for useful comments on this paper, presented at IMSFIPS2015.
Author information
Authors and Affiliations
Corresponding author
Appendices
Appendix 1
In KGB model, substituting (3.2) into (3.1) yields
where
Here i in \(d_{gij}\) denotes the attribute suffix and j the polynomial order. Thus letting
the model is reduced to a regression model;
where and .
On the other hand, the specification of the stochastic part of the DF is given by
and hence the covariance matrix of prices is
where \( \varphi _{gh} =\sum \nolimits _{j=1}^{M(g)} {\sum \nolimits _{m=1}^{M(h)} {C_g (s_{gj} )C_h (s_{hm} )f_{gh\cdot jm} } }\).
The parameter vector is estimated by generalized least squares (GLS) method and the objective function
is minimized with respect to the unknown parameters (see Kariya and Kurata 2004 for the effectiveness of GLS). First, for given \((\theta ,\rho ,\xi )\), the minimizer of this function with respect to is known to be the GLSE;
and then the marginally minimized function with substitution of is minimized with respect to \((\theta ,\rho ,\xi )\), yielding the GLSE .
Appendix 2
t \(=\) 1 | 0.95 | 0.05 | 0 | 0 | t \(=\) 11 | 0.785714 | 0.214286 | 0 | 0 |
0.021739 | 0.978261 | 0 | 0 | 0.102041 | 0.836735 | 0.061224 | 0 | ||
0 | 0.090909 | 0.909091 | 0 | 0 | 0.111111 | 0.888889 | 0 | ||
0 | 0 | 0 | 1 | 0 | 0 | 0.125 | 0.875 | ||
t \(=\) 2 | 0.974359 | 0.025641 | 0 | 0 | t \(=\) 12 | 1 | 0 | 0 | 0 |
0.020833 | 0.958333 | 0.020833 | 0 | 0.12766 | 0.87234 | 0 | 0 | ||
0 | 0.2 | 0.8 | 0 | 0 | 0.321429 | 0.678571 | 0 | ||
0 | 0 | 0 | 1 | 0 | 0 | 0.428571 | 0.571429 | ||
t \(=\) 3 | 0.923077 | 0.076923 | 0 | 0 | t \(=\) 13 | 1 | 0 | 0 | 0 |
0.040816 | 0.938776 | 0.020408 | 0 | 0.12 | 0.88 | 0 | 0 | ||
0 | 0 | 1 | 0 | 0 | 0.181818 | 0.772727 | 0.045455 | ||
0 | 0 | 0 | 1 | 0 | 0 | 0.25 | 0.75 | ||
t \(=\) 4 | 0.815789 | 0.184211 | 0 | 0 | t \(=\) 14 | 1 | 0 | 0 | 0 |
0 | 0.979592 | 0.020408 | 0 | 0 | 0.979167 | 0.020833 | 0 | ||
0 | 0 | 1 | 0 | 0 | 0.111111 | 0.833333 | 0.055556 | ||
0 | 0 | 0 | 1 | 0 | 0 | 0 | 1 | ||
t \(=\) 5 | 0.903226 | 0.096774 | 0 | 0 | t \(=\) 15 | 0.678571 | 0.321429 | 0 | 0 |
0.018182 | 0.890909 | 0.090909 | 0 | 0 | 0.897959 | 0.102041 | 0 | ||
0 | 0 | 1 | 0 | 0 | 0 | 1 | 0 | ||
0 | 0 | 0 | 1 | 0 | 0 | 0 | 1 | ||
t \(=\) 6 | 0.517241 | 0.482759 | 0 | 0 | t \(=\) 16 | 0.894737 | 0.105263 | 0 | 0 |
0 | 0.884615 | 0.115385 | 0 | 0.075472 | 0.90566 | 0.018868 | 0 | ||
0 | 0.0625 | 0.875 | 0.0625 | 0 | 0.142857 | 0.857143 | 0 | ||
0 | 0 | 0 | 1 | 0 | 0 | 0 | 1 | ||
t \(=\) 7 | 0.666667 | 0.333333 | 0 | 0 | t \(=\) 17 | 1 | 0 | 0 | 0 |
0.016393 | 0.803279 | 0.180328 | 0 | 0.037736 | 0.962264 | 0 | 0 | ||
0 | 0 | 0.9 | 0.1 | 0 | 0.105263 | 0.894737 | 0 | ||
0 | 0 | 0 | 1 | 0 | 0 | 0.2 | 0.8 | ||
t \(=\) 8 | 1 | 0 | 0 | 0 | t \(=\) 18 | 0.956522 | 0.043478 | 0 | 0 |
0.203704 | 0.777778 | 0.018519 | 0 | 0.188679 | 0.792453 | 0.018868 | 0 | ||
0 | 0.206897 | 0.793103 | 0 | 0 | 0.055556 | 0.944444 | 0 | ||
0 | 0 | 0 | 1 | 0 | 0 | 0 | 1 | ||
t \(=\) 9 | 0.681818 | 0.318182 | 0 | 0 | t \(=\) 19 | 0.96875 | 0.03125 | 0 | 0 |
0.020833 | 0.875 | 0.104167 | 0 | 0.181818 | 0.795455 | 0.022727 | 0 | ||
0 | 0.041667 | 0.833333 | 0.125 | 0 | 0.222222 | 0.777778 | 0 | ||
0 | 0 | 0.25 | 0.75 | 0 | 0 | 0 | 1 | ||
t \(=\) 10 | 0.8125 | 0.1875 | 0 | 0 | t \(=\) 20 | 1 | 0 | 0 | 0 |
0.02 | 0.9 | 0.08 | 0 | 0.175 | 0.825 | 0 | 0 | ||
0 | 0.038462 | 0.884615 | 0.076923 | 0 | 0.266667 | 0.733333 | 0 | ||
0 | 0 | 0 | 1 | 0 | 0 | 0.5 | 0.5 |
Appendix 3
Assuming the covariance structure in (6.2) for the stochastic discount function yields
where \(\lambda _{kl} \) is the same as the one in KGB model.
However a significant difference is that this covariance matrix depends on the unknown regression parameters to be estimated. Of course our objective function to be minimized is of the same form as KGB model;
where the covariance matrix is replaced by the matrix (\(\sigma ^{2}\lambda _{kl} \varphi _{kl} )\) in (10.1). To get an approximate minimize, the GLS is applied to (6.2) in a repeated manner. In fact, we set as the initial value in (10.2) to obtain the first GLS estimate from (6.2) and (10.1). Next we insert into (10.2) and get the second GLS estimate from (6.2) and (10.1). Repeating this procedure five times yields our GLS estimate the coefficients of the TSDP in (6.4). The minimized value is . The parameter \(\theta \) is always estimated as 0. For the theoretical background for the GLS, the readers are referred to Kariya and Kurata (2004).
Rights and permissions
About this article
Cite this article
Kariya, T., Tanokura, Y., Takada, H. et al. Measuring Credit Risk of Individual Corporate Bonds in US Energy Sector. Asia-Pac Financ Markets 23, 229–262 (2016). https://doi.org/10.1007/s10690-016-9217-7
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10690-016-9217-7