Synopsis: Adopting a stricter measure of the CCC sector does not alter our conclusion that for the past few months, the lowest credit-quality tier has represented the least overvalued segment of the high-yield market. The high-yield asset class as a whole is extremely overvalued at present. On the positive side, the covenant quality of new issues has been improving since the fall.
Purifying the CCC segment to determine relative value
In conjunction with this update of our rating-specific fair-value analysis, we address a concern raised by some readers about measurement of the CCC spread. As detailed in “Valuation and short-run performance” (Feb. 19, 2013), our model explains historical variance in the option-adjusted spreads of the BofA Merrill Lynch U.S. High Yield Master II Index rating subdivisions as a function of economic conditions, credit availability, the speculative-grade default rate, and Treasury yields.
The bottom-tier index (BAML Ticker: H0A3) consists primarily of CCCs, but also includes CCs and Cs. Both in the latest month and in the historical period used to create the multiple regression formula by which we estimate fair value, the H0A3 spread exceeds the pure CCC spread. This difference could conceivably lead to a false comparison of relative value between CCCs and the other rating categories.
To address this concern, we created a new time series for the pure CCC spread, consisting of 68 quarterly observations (December 1996 to December 2013). In each quarter we calculated a market-value-weighted OAS for the CCCs only. We then applied the above-described FridsonVision Fair Value Model to the quarterly data. The resulting model (see note 1) explains 76% of the variance in the pure CCC spread. This level of explanatory power (R-squared) compares favorably with the corresponding figures of 78%, 80%, and 74% for H0A1 (BB), H0A2 (B), and H0A3 (CCC/CCC/C), respectively.
Fair-valuation output for Jan. 31 appears in the chart below. The bottom line is that whether defined as H0A3 or our newly created Pure CCC index, the bottom tier of the speculative-grade asset class continues – as in recent months – to be the least overvalued.
This is true even in absolute terms for the CCC/CC/C group. Its actual-versus-fair-value gap is less than that of the B category (145 versus 182 bps). On the other hand, our Pure CCC group’s gap versus fair value is five basis points greater the B group’s (187 versus 182). These gaps must be normalized, however, for the differences in scale of fair-value OAS for the various rating categories.
To understand why, suppose that on a given date, the fair-value spread for AA corporates is 100 bps. If the actual spread on that date is 200 bps, AA corporates are surely cheaper than CCCs, with a fair-value spread of 1,000 bps on the same date and an actual spread also greater by 100 bps, i.e., 1,100. Surely, the AAs are cheaper, at a spread 100% greater than fair value, than CCCs, which are a mere 10% wider than fair value.
Following that logic, we show in the column labeled “Difference as a % of fair value” that the CCC/CC/C group is the least overvalued (-15.8%) and the B group is the most overvalued (-31.1%). Pure CCCs, at -22.1%, are more overvalued than the CCC/CC/C category, but still less overvalued than either BBs or Bs. An alternative approach to normalizing the actual-minus-fair-value gaps, measuring the number of standard deviations of overvaluation, produces the same result – namely, the bottom tier is the cheapest portion of the high-yield market at present.
As a caveat, short-term relative returns tend to be a function of changes in investors’ market tolerance, rather than valuation. If risk premiums increase sharply, CCCs will almost certainly underperform BBs and Bs, even if they are relatively undervalued at the outset. Valuation is not a tool for short-term traders, but rather a guide for value investors who aim to benefit over the long run by perennially holding intrinsically underpriced assets.
Going forward, our monthly valuation update will include fair value and actual spreads for the Pure CCC index as well as BAML’s H0A3 index.
High-yield is extremely rich
The chart below updates the fair-value analysis of the high-yield asset class, initially described in “Determining fair value for the high-yield market” (Nov. 13, 2012). Taking into account total risk (not just default risk, as primitive “breakeven” models do), fair value for the High Yield Master II is 557 bps at present.
The current fair-value spread is down from 568 bps one month earlier. An improvement in credit availability offset unexpected, substantial drops in Industrial Production and Capacity Utilization. (Unusually harsh winter weather probably contributed to those declines.) The default rate declined slightly, while the five-year Treasury yield, which is inversely correlated with the spread, dropped by 16 bps.
On Feb. 14 the actual high-yield OAS was 403 bps, or 1.2 standard deviations less than fair value. We define a shortfall of one standard deviation or more as an extreme overvaluation. High-yield has historically underperformed Treasuries in 12-month periods following extreme overvaluations. For the record, the Jan. 31 actual OAS was 421 bps, representing a divergence of 1.0 standard deviations.
Covenant quality improved in January
Covenant quality on new high-yield issues improved in January, continuing an upward trend from the recent low point in the fall of 2013. The chart below updates analysis introduced in “Covenant quality decline reexamined” (Oct. 1, 2013). The Moody’s version of the covenant-quality trend is shown by the dotted line, while the solid line depicts the FridsonVision version. Both are based on Moody’s scores for the individual high-yield issues that come to market each month.
The methodological difference is that we filter out the impact of month-to-month changes in the ratings mix of new high-yield issues. CCC issues on average have stronger covenants than Bs, which on average have stronger covenants than BBs. If, in a given month, the balance of negotiating power between issuers and investors does not shift, but CCCs happen to represent a greater share of issuance than in the preceding month, the Moody’s series will show an improvement in covenant quality, but ours will (correctly) show no such change.
Our series indicates an improvement in the CQ trend to 3.89 in January, from 4.03 in December and from a worst point of 4.19 in September 2013. The Moody’s series shows a worst point of 4.26 in October 2013 and a latest-month improvement to 3.84, from 4.00 in December. By either measure, the rockier conditions in the high-yield market of late have weakened the hand of issuers, enabling investors to obtain somewhat stronger covenants than in the fall.
Martin Fridson, CFA
CEO, FridsonVision LLC
Research assistance by Yinqiao Yin and Ruili Liu
1. The multiple regression formula is: