Empirical Studies of Credit Spreads in the European High Yield Market

Mikkel Lindkvist Nielsen & Christina Marie Lie

Student thesis: Master thesis

Abstract

Using data from the Danish manager Capital Four Management, we study the spread determinants of two types of European High Yield bonds. We firstly test the possible use of bond characteristics and accounting variables as a way of explaining spreads (and thereby pricing) of standard high yield bonds using OLS regression techniques and a set of 518 observations. We secondly discuss the recent increase in the issuance of contingent convertible (CoCo) bonds in the European High Yield Market and attempt to estimate the spread determinants of these with a focus on CoCo specific risk determinants that are special to these bonds. On an overall level, our models are quite successful in the sense that they can explain most of the variation in spreads. Our model of standard high yield bonds are able to explain up to 73% of the variation in spreads. We furthermore show that both bond characteristics and accounting variables are useful tools in a spread model. The bond characteristics alone can explain an impressive 46% of the variation in spreads, while the accounting variables alone explain 35% of the variation. Our model of spreads of CoCo bonds turns out to be even more effective and is able to explain 74% of the variation in spreads. From our sample of standard high yield bonds we are able to show that credit risk indeed have an effect on the spread. Using accounting variables as proxies for credit risk, we find that most of these variables behave as one would expect them to according theory. We also find that bond characteristics can have a significant effect on spreads, as these also affect credit risk. Lastly, our results show a negative effect of issue size, which could indicate that liquidity risk is also a factor in the spread determination of high yield bonds, such that smaller issues have higher liquidity risk, and hence higher spreads. Our analysis of CoCo bonds shows that the spreads of CoCo bonds do not only depend on the overall credit risk of the bank, but also on CoCo specific risk that are only present in CoCo bonds. Our findings first of all suggest that CoCo bonds become more risky (trade at a higher spread), the closer the bank’s capital ratio is to the trigger level. We furthermore find that the spread of a CoCo bond depends negatively on the amount of capital a bank is able to generate internally (as opposed to using the capital markets), as this lowers trigger risk. Unfortunately, we are not able to show that coupon deferral risk has an effect on CoCo spreads, although this should be the case. We suggest that this is due to difficulties in measuring this risk. Lastly, we find that CoCo bonds issued in USD have a significant lower spread than CoCo bonds issued in EUR. This effect is definitely not non-negligible as the average discount on USD CoCo bonds is 105 bps. We discuss various explanations for this finding.

EducationsMSc in Applied Economics and Finance, (Graduate Programme) Final Thesis
LanguageEnglish
Publication date2015
Number of pages131