Fixed Income Fund Selection in the Danish Covered Bond Market

Morten Guldbrandtsen Gilder & Mathias Stauning

Student thesis: Master thesis

Abstract

This thesis investigates fund selection in the Danish covered bond market. This is done by creating a proprietary dataset compiled from both financial databases and cover pool reports from issuers of covered bonds.
The data is considered as four distinct sections. These sections are macroeconomic, fund characteristics, security specific variables and cover pool details. The factors in each section are first described theoretically, then analyzed empirically and finally discussed in order to determine if the factor is economically significant. As the analysis progresses, strong factors are kept while weaker variables are excluded from the selection model moving forward. The macroeconomic section finds that the sample period is dominated by falling interest rates and that performance above the benchmark has been positively affected by this. For the fund characteristics, it is discovered that expense ratios help describe fund performance in a non-linear manner, with the implication that the optimal expense ratio is slightly above the minimum observation. Furthermore, it is found that funds that are highly exposed to interest rate risk through duration have historically yielded higher returns than the benchmark. This is due to the observed interest rate development. Examination of the cover pool data finds that funds exposed to relatively higher default risk has performed well, as no bond defaults have been observed in the sample period. Relating to this, it is also found that exposure to riskier types of collateral has driven outperformance. Finally, investing in cover pools with an overweight in large loans has led to underperformance, as it is argued that the falling interest rates have led to significant prepayments in this category of loan size. The result is a well-founded model that incorporates only the most significant factors.
It is noted that the final model is affected by omitted variable bias and heteroscedastic errors, which means that the model implications have to be interpreted accordingly. Testing the model over several business cycles finds that only exposure to default risk is robust over all cycles. However, due to the limitation of the economic environment of the sample period, the forward-looking capabilities of the model are restricted. Further research into this subject is encouraged once data based on other interest- and default paradigms is available, in order to validate and improve on the results found in this thesis.

EducationsMSc in Finance and Investments, (Graduate Programme) Final Thesis
LanguageEnglish
Publication date2017
Number of pages167
SupervisorsSøren Plesner