Determinants of Performance for Mortgage-Backed Securities Funds

Article excerpt

This study examines the performance of the 31 mortgage-backed securities (MBS) mutual funds that existed throughout the January 1987-June 1995 period. There are three reasons for investigating the performance of MBS funds. First, the MBS market is the second largest sector of the United States debt market behind the U.S. Treasury sector (Federal Reserve Bulletin, October 1995), and, thus, represents a significant asset class. For many investors, however, the only way of investing in mortgage securities is through mutual funds specializing in mortgage-related securities. The recent growth of MBS funds is indicative of their popularity. For instance, at the end of 1986, there were 31 MBS funds with $40 billion in assets. By mid-1995, there were 133 MBS funds holding $56.71 billion in assets (Morningstar Mutual Funds, July 1995).(1) Despite this rapid growth, surprisingly little research has examined the investment performance of MBS funds.

Second, the choice of the MBS fund benchmark is important both for the management of MBS funds and for the evaluation of fund performance. Funds routinely measure their performance relative to benchmarks. Thus, it is important to determine which MBS benchmark is more appropriate for evaluating MBS mutual funds. Both the Lehman Brothers and Salomon Brothers MBS market indexes are examined. Prior studies of domestic bond fund performance have utilized both single and multi-index models. In particular, Kritzman (1983), Cornell and Green (1991) and Blake, Elton and Gruber (1993) find that domestic bond funds generally do not outperform various bond benchmarks.

Third, the features associated with the better performing funds are identified by running tests to explain performance. Specifically, since MBS funds also invest in treasury bonds, the effects of fund exposure to interest rate risk inherent in the government bond market are examined. The effects of prepayment risk that might not already be reflected in the MBS market benchmark are also examined. Tests on fund characteristics are run such as load, expenses, portfolio turnover, management fees, asset size, duration and whether the fund invests in CMOs or other derivatives. The separate effects of the fund manager's security selection and market timing are also measured. Last, tests are performed to determine if performance is consistent across the sample period or if performance is driven by a few unusually good or bad months.

The findings indicate that, on average, MBS funds underperform the MBS market index. The underperformance of the funds relative to the MBS market index is not the result of differential exposure to government bond market risk or to prepayment risk not already reflected in the MBS market benchmark. Poor securities selection and timing contribute significantly to the underperformance of the MBS funds relative to the MBS benchmark. Funds with high expenses are likely to underperform the MBS benchmark. The size of the fund, the fund's duration, portfolio turnover, load and management fees do not affect performance in a material way. Nor does the decision to invest in CMO or in other derivatives. Upon closer examination, the underperformance is found to be concentrated in a few particularly bad months. After deleting the outlier months, the MBS funds match their benchmark return. Further testing shows that the MBS mutual funds underperform the MBS benchmarks during months of rising interest rates. Performance of the funds during months of falling interest rates matches the MBS benchmark. The remainder of this article consists of four sections. The next section discusses the data. The third presents the methodology. The fourth section discusses the empirical findings and the last section is the conclusion.


The monthly returns for all MBS mutual funds that existed throughout the January 1987-June 1995 period and that were included in the CDA Investment Technologies, Inc. database are examined. …