Just because junk-bond ETFs trade well does not mean they track their asset class well.
The popularity of indexing has become indiscriminate. Some asset classes are not amenable to indexing, and some exchange-traded funds should be avoided for that--especially junk-bond ETFs. These products fail on two levels: Their underlying asset class may be systematically overpriced, failing to offer returns commensurate with their risks, and their price-indifferent trading imposes enormous hidden costs on investors.
Illusory Returns, Real Risk
At first glance, the historical record contradicts me. Over the past 26 years, junk bonds, proxied by the BoAML U.S. High Yield Master II Index, have returned 8.8% annualized with an 8.6% standard deviation, for a 0.58 Sharpe ratio. Most investors would envy such a record. However, much of the return came from decades of declining interest rates, which provided a big capital appreciation boost. Junk bonds look less impressive against duration-matched Treasuries, outperforming by 200 basis points and with much higher volatility. Not bad, but much of the return advantage is illusory. To get a sense of how difficult it was to obtain the index's returns, look at junk-bond mutual funds. Of the 27 high-yield funds that have survived since 1986, the starting year of the high-yield index, not one earned a higher Sharpe ratio than the index, and only one posted higher absolute returns. These are the winners; the ones that died off posted much worse returns.
- source: Morningstar Analysts
Mutual funds had a hard time matching index returns owing to the illiquidity of the market. Many junk bonds trade irregularly, so index prices are often estimated using "matrix pricing" models, which produce interpolated prices based on a handful of reference bonds. These prices are often impossible for investors to obtain. If so few funds can get close to the index's returns, then the index's returns are fiction. We get closer to reality with liquidity-screened indexes, such as the iBoxx Liquid High Yield Index.
Since its 1999 inception, iBoxx's index has trailed the broad junk-bond index by 1.72%, eliminating most of the asset class' excess returns. This finding is consistent with previous studies showing that much of the excess returns from less-liquid asset classes come from the least-liquid tranches. According to the Ibbotson SBBI 2012 Yearbook, over the period 1972 to 2011, the least-liquid quartile of small-cap U.S. stocks drove much of small caps' outperformance; the most-liquid small caps actually lagged the most liquid large caps by 6.5% annualized.
With much of the illiquidity premium excised from investable junk-bond indexes, you're left with an awful asset class that has provided scant return above Treasuries, more volatility, and, worst of all, inferior diversification. Junk-bond ETFs during the financial crisis experienced 30%-plus losses, whereas duration-matched Treasuries shot up in the flight to quality.
The poor compensation for liquidity risk junk-bond investors have received suggests the market has historically been overpriced. Of course, poor realized returns don't condemn an asset class or strategy, even over a period of as long as three decades. But they are very suggestive when the data are considered in context.
1) Credit risk has been poorly compensated. From 1926 to 2012, long-term corporates have returned less than 0.40% annualized over long-term government bonds, with most of the excess returns occurring during the 1930s. Over the past 50 years, credit risk in long corporate bonds has provided close to zero excess return over Treasuries.