Credit Risk in Corporations: To What Extent can Structural Pricing Models be Improved?

Mathias Nørmark Thrane & Sebastian Billum-Jensen

Student thesis: Master thesis


This paper examines if the structural framework for credit risk pricing can be improved upon by incorporating accounting based information. Structural models of credit risk pricing has been subject to criticism due to its imprecise prediction of credit spreads. In particular, the model has been found to underpredict credit spreads. In the past decade, scholars have been trying to estimate whether accounting information can play an incremental role in the prediction of credit spreads. So far, scholars have yet to reach a general and/or shared consensus on this questionable matter. Common for these previous studies, only a limited number of accounting variables have been considered in conjunction with basic structural models. This paper follows the basic methodological approach suggested by earlier research, but distinguish itself by making use of a more extensive set of accounting variables as well as a more comprehensive structural model. The structural framework of credit risk pricing is initially evaluated for its predictive power. The evaluation suggest that the model developed by Longstaff and Schwartz (1995) performs best. This model, is then used as the reference point for improving the structural framework. By developing a pure accounting model that accounts for time period and firm fixed effects, the paper finds that a number of accounting measures are able to explain prices of credit risk. To examine the combined information, a hybrid model that incorporates both accounting and structural model information is then developed. The results obtained by the hybrid model shows that accounting measures for solvency and size are already accounted for in the structural framework. However, the results suggest that profitability, liquidity and growth ratios, all provide additional information that can improve the structural model. In order to ensure that the improved power of the hybrid model is not solely attributable to the control of time period and firm fixed effects, a robustness test is made where these effects are incorporated into the model of Longstaff and Schwartz (1995). The robustness test validates that these three accounting categories can improve the structural framework. Furthermore, it finds that incorporating these three accounting categories into the model, allows it to explain 5% more of the variance in observed credit spreads.

EducationsMSc in Applied Economics and Finance, (Graduate Programme) Final ThesisMSc in Finance and Investments, (Graduate Programme) Final Thesis
Publication date2018
Number of pages135
SupervisorsJeppe Christoffersen