In a perfect world, a reasonable person would hope that the net asset value (NAV) of an exchange traded fund (ETF) would equal the index return minus the expense ratio. The difference is called the tracking error. On this basis, it's interesting when a fund exceeds that benchmark and raises the question: how do they do that?
For the three years ending June 14th, 2011, the Barclays Aggregate Bond Index (NYSEARCA:AGG) has earned 6.95 percent according to Morningstar. The Vanguard Total Bond (NYSEARCA:BND) fund charges an expense ratio of 0.11 percent, so a reasonable investor would expect the NAV to have gained by 6.83 percent (that extra basis point is due to compounding: (1.0695 / 1.0011)-1 = 0.0683 = 6.83 percent).
According to Morningstar, however, the NAV for BND was actually 6.94 percent – one basis point away from the index! What explains this difference?
Normally, these differences are chalked up to trading techniques. This is true, but also not particularly descriptive, so it might be interesting to investigate one strategy more closely.
Let's start with the fact that while Vanguard seeks to fully replicate the AGG, there are some practical limitations that make 100 percent perfect replication of the 7,989 bonds in the index impossible.
Vanguard has almost 5,000 individual bond positions, which is short of the index, but well more than any of its competitors tracking this index that mostly have less than 1,000 bonds.
If anyone could fully replicate, it would be Vanguard, thanks to their patented share class structure. To the extent that they can't fully replicate, they have to make some optimizing decisions.
Consider the corporate bond sector allocation, which may be one of the hardest areas to fully replicate due to the illiquidity of some of the individual bonds. Here, Vanguard is able to replicate approximately 90 percent of the sector, but can't efficiently replicate 100 percent.
For the remaining 10 percent or so that can't be replicated, Vanguard uses their team of credit analysts to add a one or two basis point overweight to issues that the analysts view as improving and subtracts one or two basis points to underweight names that they believe aren't as attractive.
These slight over and underweights do not change the character of the fund – the curve still matches, the credit rating and quality still matches, the sectors still match, etc. Even with these small optimization decisions, there are no active tilts in the fund.
Vanguard says that their risk models indicate that on an ex ante basis, the tracking error will be extremely tight, somewhere around five basis points. Over the past three years, it's been better than their models would have predicted and been in favor of the fund holders. Nice.
Disclosure: I am long BND. The views expressed do not necessarily represent the views of Acropolis Investment Management, LLC. or its members. Clients of Acropolis own BND as a core holding and we are regularly buying and selling BND for clients as part of our portfolio rebalancing.