Predicting Bond Fund Returns

There is disagreement among prior studies as to whether bond fund returns can be predicted. Reichenstein (1999) suggests that these differences in conclusions are likely due to differences in the samples. When the sample contains funds across investment styles, multiple studies conclude that bond fund returns are not predictable. But Reichenstein concludes that when the sample contains funds with similar investment styles, expense ratios can partially predict bond funds ' relative returns. For example, suppose a sample contains high-grade funds with short- to long-term durations. In periods when interest rates rise, short-duration funds will tend to be the better performers. In periods when interest rates fall, long-duration funds will tend to be the better performers. If interest rates either rise in two consecutive periods or fall in two consecutive periods, then funds with above-average returns in the first period are likely to produce above-average returns in the second period. However, when either interest rates rise and then fall or fall and then rise in the two periods, then funds with above-average returns in the first period will generally produce below-average returns in the second. Since returns to bearing interest rate risk are not consistent across periods, relative bond returns should not be predictable in samples tha