Bond funds' past performance isn't necessarily indicative of skill.
A version of this article appeared in the September issue of Morningstar ETFInvestor. Download a complimentary copy of ETFInvestor here.
Bond investors require compensation for bearing greater credit and interest-rate risk through higher expected returns, or yields. Fixed-income managers can--and often do--boost returns by taking more credit or interest-rate risk than their benchmarks, but that does not necessarily require skill. Investors can do that on their own with low-cost index funds, so it is hard to justify paying high fees for performance that is simply compensation for risk. And while risk taking can pay off in one period, it may lead to underperformance in another.
Therefore, it is important to distinguish between returns that managers generate through risk-taking and skill (including the skill involved in timing interest and credit risk). We can accomplish this with a factor regression analysis.
To illustrate, consider a simple model that seeks to explain PIMCO Total Return'sPTTRX return as a function of interest-rate risk. To construct this model, we first subtract the returns on short-term Treasury bills from the fund's returns to isolate the extra return investors get for putting their money at risk. We could also subtract the return on Treasuries from the return on the Barclays Intermediate U.S. Treasury Index to determine the payoff the market offers for intermediate-term interest-rate risk (independent of credit risk). This is an interest-rate factor.
The chart below shows the relationship between the fund's excess monthly returns (over Treasuries) and the excess returns on the Barclays Intermediate U.S. Treasury Index over the trailing 10 years through July 2015.
Not surprisingly, there appears to be a positive relationship between the returns on the intermediate-term Treasury bond index and PIMCO Total Return. The slope of the best-fit line through the data measures how much interest-rate risk the fund is taking. A slope of less than 1.0 indicates the fund is taking less interest-rate risk than the index. In this case, the slope is 0.76, indicating that each percentage point return to intermediate-term interest-rate risk tended to contribute 76 basis points to the fund's return. More formally, this is the interest-rate factor coefficient, or beta. The point where the line intersects the y-axis (the y-intercept) captures the return from skill. This figure is better known as alpha. If a fund is not taking more risk (as defined in the model) than Treasuries, it should not generate higher returns. Therefore, if the intercept is significantly positive, either the model is incomplete or the manager is skilled. In this case, the intercept is positive, but it is not statistically significant--meaning that this observation could be due to chance.
This simple model only explains 31% of the variance in the fund's returns, suggesting that there is a lot that it does not capture. The model also has an important limitation--it assumes that there is a linear relationship between interest-rate risk and return. But the yield curve is rarely a straight line. Consequently, the model may over- or understate the return to skill.