Not all bond ETFs use credit ratings to determine what's included in the underlying portfolio, although most do. As we highlighted in our feature article from a couple of days ago (“A Messy Break-Up?”), EuroMTS doesn’t set a minimum ratings requirement for eurozone government bond issuers when compiling its most widely used eurozone benchmarks; by contrast, Markit, Europe’s leading provider of benchmarks for bond ETFs, sets a requirement that issuers are rated “investment grade” by at least one of the three major agencies (Standard and Poor’s, Moody’s and Fitch) for any of its iBoxx Euro index family members, including the widely followed Eurozone Euro Sovereigns index. When more than one agency of these three rates an issuer, then Markit uses an average rating.
Illustrating the difference between the different index methodologies, Greece was ejected from the iBoxx indices at the end of June after its downgrade by Moody’s to a “junk” rating of Ba1 on June 14, after an equivalent move by Standard and Poor’s in April (the two ratings, combined, meant that Greece’s average rating was now below investment grade). However, Greece remains in the EuroMTS broad and global indices, albeit at a modest weighting of around 3.5%.
Performance-wise, a comparison of two of Europe’s largest diversified government bond ETFs since the day of Moody’s Greek downgrade in June, one following an index that continues to include Greece, one tracking a benchmark that excludes the country, doesn’t tell us much. Lyxor’s ETF EuroMTS Global (which includes Greece) is down 0.04% over the period since June 14, while db x-trackers’ iBoxx Euro Sovereigns Eurozone Total Return ETF (which excludes it) is up 0.46% - a small difference. In fact, much of the adverse performance of Greek bonds had occurred prior to June’s downgrade, while the more recent sell-offs in some of the other eurozone bond markets (Ireland, Spain, Portugal) are reflected in both these ETFs.
Since the bond markets of eurozone member governments started to diverge in performance in 2008 (something that is beautifully illustrated in a chart from a feature we wrote in March 2009 – see “Sovereign Default Risks On The Rise”), index providers and ETF issuers have made an attempt to allow investors to segregate the overall market on the basis of credit quality. Both EuroMTS and iBoxx now offer AAA-only indices and non-AAA indices. Essentially, investors have a choice between the highest quality debt (which should come with a lower yield), or the riskier stuff and more income.
However, there are several problems with a ratings-based approach to constructing bond indices.
The first criticism is a similar one to that made by opponents of capitalisation-based equity index investing. Investors are forced to buy in at a high or, more importantly, sell at a low when an issuer breaches the minimum ratings requirement and leaves the index, in the same way as cap-weighted equity index investors have to sell out of a stock when it’s already underperformed.
Another criticism is that the ratings agencies are too slow to react to deteriorating creditworthiness, and when they do react, cuts tend to come in one fell swoop. In Greece’s case, the ratings cut to junk by Moody’s in June was one of four “notches” in one go, for example (from A3 to Ba1). One minute your issuer is in the index, with some cushion for error, the next it’s suddenly through the floor and out of the benchmark.
With Ireland's rating brought down yesterday by Standard and Poor’s from AA- to A (two notches on the scale), the issuer is now only four grades above junk status (Moody’s, by the way, still has Ireland at Aa2, three levels above the equivalent S&P rating). Since pressure on government finances is increasing everywhere, you can expect several other issuers to face downgrades, risking the sudden removal of more bonds from ratings-based benchmarks.
Finally, and perhaps worst, ratings methodologies are not consistent across the markets. It’s easy to find lower-rated issuers with bonds offering a higher yield (and higher risk, theoretically) than higher-rated ones, something that doesn’t make sense.
The worst abuse of the ratings system, of course, was the widespread grade inflation in structured finance securities during the credit bubble, with the AAA label incorrectly attached to bonds that both risked and then produced a severe loss of capital.
Even though the structured finance market is now comatose, contradictions abound in the way simpler (bullet) bonds are rated. For example, Russia (rated Ba2 by Moody’s) is paying a yield of around 4.75% on its ten-year dollar debt, while Aa2-rated Ireland (that’s nine credit ratings better than Russia, according to Moody’s) has a current yield of 9.15% on its ten year euro-denominated bonds. Something’s badly wrong here. You need to adjust (slightly) from a dollar yield curve to one in euro when making this comparison, but a glaring inconsistency remains.
What’s the solution? Steer clear of a ratings-based index methodology altogether, I’d suggest. Whether there’s a way of using market-based credit spreads (via sovereign credit default swaps) to construct passive bond portfolios is a question I’ll leave to the index and ETF product development experts. But the current way of putting many bond ETFs together seems deeply flawed.