Is The Bond Index Broken?
December 1, 2015
by Michael Edesess
When equity investors want to compare their performance to a broad U.S. index, they use one representing the whole equity market like the Wilshire 5000, or the S&P 500, which is a little less broad. When they want to compare fixed income performance to a benchmark, they usually use a bond index like the Barclays Aggregate.
Several criticisms – I counted three – have been leveled at the bond index recently. I explored these critiques in a wide-ranging conversation with John C. Bogle, renowned founder of The Vanguard Group. My conclusion is that two of these three criticisms are inconsequential or mistaken while the third is meaningful and significant. And yet, one of the discarded criticisms highlights a question I have been curious to answer – and even Bogle himself, one of our most important big-picture thinkers, was not steeped enough in the minute details of index fund investing to answer it.
The three criticisms of the bond index
Inquiries made to people with long experience in the investment field identified three criticisms that have been leveled at the bond index:
- It overweights heavily indebted issuers.
- It may be double-counting some fixed income securities.
- It is not representative of most investors’ fixed income portfolios; this was expressed by Bogle himself in two conversations with Morningstar.com interviewer Christine Benz.
Let us consider these criticisms in turn.
Does the bond index overweight heavily indebted issuers?
This sounds at first like an extension to fixed income of the canard that a capitalization-weighted equity index overweights large companies – that is, those companies with larger amounts of stock outstanding.
However, perhaps it is a little different in the bond case. Dan diBartolomeo of Northfield Information Services said it was related to the concern that many bonds are relatively illiquid and therefore are not priced regularly. As a result their returns tend to be autocorrelated – a consequence, presumably, of the fact that when a current market price is unavailable the default price is the last price the bond was traded at, whenever that was.
If this is true, then the size-weighting of the bond in the index may be inaccurate; those companies with overpriced bonds may be able to force additional debt on indexers who are obliged to accept it at an excessive price.
Bogle thought that such substantial mispricing of bonds in, for example, Vanguard bond index funds was extremely unlikely. If a bond were substantially mispriced it would be obvious from its yield, which would be out of line with that of similar bonds. Vanguard would pursue the pricing source to get to the bottom of the discrepancy. (This pursuit might, in the end, result in a bond price that was re-marked not to market but to model, perhaps a less-than-ideal solution but one that would be less likely to result in a price distortion.)
If the bonds are all efficiently priced – or at least as close to efficient as possible – then there would be no reason to believe that larger issuers are cutting a better deal from index investors or are more risky than smaller issuers (as there is also no reason to believe the same for equities). If an issuer becomes more risky then the process of issuing bonds would become self-limiting. The bonds would have to be issued at higher and higher interest rates and would eventually become unattractive for the issuer, or wouldn’t be bought. In other words, an efficient market would ensure that an issuer wouldn’t issue so many bonds that they would become an unattractive investment. And when the market is inefficient the issue is as likely to be underpriced as overpriced.
Does the bond index double-count?
This criticism is that the bond index could contain a bond while also containing the same bond as part of an asset-backed security. It seems highly unlikely – though possible – that this could be true.
The Barclays U.S. Aggregate Index composition is shown in the following chart:
Source: Barclays Risk Analytics and Index Solutions
The securitized portion consists, according to Barclays, of “MBS (agency fixed-rate and hybrid ARM pass-throughs), ABS and CMBS (agency and non-agency).”
Thus, the bulk of the securitized portion of the index apparently consists of mortgage-backed securities. Since no mortgages are included elsewhere in the index, these could not be a source of double-counting. The only other possibility is the ABS. It is conceivable that these could include packages consisting, at least partially, of securitized corporate debt, some of which is also included in the index as individual corporate bonds. Nothing specified in Barclays’ document describing the U.S. Aggregate Index (downloadable here) explicitly rules out this possibility. Nevertheless, it is likely that the effect would be negligible even if Barclays had not made an adequate effort to avoid double-counting.