Matt’s call for a better bond index is just the tip of the iceberg.
I don’t disagree that the big bond indexes are busted. The problem is that (despite what you read on the Web sites) the big bond indexes share far more in common with the Dow than they do with the S&P 500.
The Lehman Aggregate and its various offshoots (and Barclays Capital rebrandings) were really designed to represent what was happening in bonds, not to be a buy-list. This isn’t news to any bond investor really—none of the big bond ETFs owns anything like the actual composition of the indexes they purport to track. The performance of the iShares and SPDR versions of the Barclays Aggregate (NYSEArca: AGG and NYSEArca: LAG, respectively) can at best be considered correlated, but hardly perfect matches.
And for obvious reasons. AGG holds 247 out of a potential 10,000 bonds in the actual index, LAG just 191. And while there’s obviously a lot of overlap, little differences matter in fixed income. Checking yesterday, AGG’s largest holding was a 4.7 percent allocation to a 30-year AAA-rated government bond with a 4.5 percent coupon, while LAG had 8.13 percent of its assets in a similarly rated 30-year bond with a 5.5 percent coupon. LAG counts a Citigroup bond due in 2011 in its top 10 holdings, whereas AGG’s Citi exposure is measured in basis points.
While neither fund is likely to blow up over these variations, they are going to perform differently.
All this noise is why the conventional wisdom has so long remained that bonds were the one corner of the market where active management made sense. Let’s face it, the decision to have Citi at 2.57 percent of the LAG portfolio is fundamentally an active call, regardless of the algorithm or policy that drove it. It’s also a stark contrast to the Vanguard model, where the shared portfolio with the institutional and retail fund products means there’s actually a 12,000 bond portfolio underneath the Vanguard Total Bond Market ETF (NYSEArca: BND). It helps that the ETF invests alongside all the other Vanguard share classes, and thus has 23 years of accumulation history and $63 billion to play with.
The real issue here is that the idea of monolithic bond indexes just doesn’t work. There may only be one security called “MSFT” in the S&P 500, but how many dozens of flavors of Citigroup bonds are still floating out there? I count 83. Do the math on the number of potential issuers out there and it’s a nearly intractable problem.
Simply calling on the spirit of Rob Arnott to invent a better index isn’t the answer. Arnott and his posse have already done the heavy lifting here: They published a paper in 2008 which did exactly what you want them to do—apply all the fundamentals to creating a new index. They found that while the fundamental indexes they posit do backtest better than the simple cap-weighting systems currently used, they have a huge matching problem: Only 65 percent of corporate bonds match up well with a corporate entity on which fundamental valuation techniques can be used.
That doesn’t mean it doesn’t work—it just means that a fundamental corporate bond index brings a pile of flaws all its own to the table, yet another source of noise in an already noisy marketplace.