Measuring and predicting bond fund performance : an empirical study of the Norwegian market
MetadataVis full innførsel
- Master Thesis 
This paper concentrates on the performance of Norwegian bond funds by measuring the risk-adjusted return (alphas) and examining the predictive power of several fund characteristics. We use daily returns both gross and net of expenses on 18 actively managed corporate funds between October 2006 to September 2016. In the first part, the performance is measured by employing a single-index model and several multi-factor models over the full ten-year period. We find that about 70% of bond funds have been able to generate significant abnormal returns gross of expenses. After adjusting for expenses, only about 30% of the funds generate significant out-performance. Moreover, there is not a single fund exhibiting a significant negative performance. A non-constrained multi-factor model that captures the term and default premium best describes the return variation of these funds. In the second part, we test whether abnormal performance can be predicted while accounting for relevant characteristics which can impact the future performance. The analysis is conducted over twenty half-year periods using three multifactor models. We find evidence that persistence in abnormal return during the current half year period is a significant predictor of abnormal performance over the next half year period. Our analysis reveals approximately a third of the risk-adjusted return over the current period carries forward to the next period. There is an insignificant relationship for all other factors. The result is found to be robust across all multi-factor models.