Projections do make sense at 5-year range
Whatever the polar opposite of "You can take it to the bank" is, you can apply it to those high-yield total return projections for 2021 that you may have seen in the past few weeks.
Yes, one of them — or at least one of the revised forecasts that the prognosticators may issue a few months from now — may happen to approximate the actual return for the year. It is also true that, through sheer luck, somebody will buy the winning ticket at the next charity raffle you attend. Being right one time, however, is no proof of skill in forecasting the 12-month total return, even if the forecaster is trying really hard. Trying really hard, in this context, means doing something more than devising specious arguments to justify a forecast of "the current yield plus a couple of points from spread-tightening," simply because that is the least optimistic forecast that the forecaster's employer will allow to go out to clients (see note 1).
It is a safe bet that nobody published a forecast anywhere close to negative 26.39%, the actual 2008 total return of the ICE BofA U.S. High Yield Index. Neither is it likely that anyone was bold enough to predict that the index would return something in the vicinity of 57.52% one year later. Those two percentages, respectively, the lowest and highest in the ICE BofA U.S. High Yield Index's history, make a point that is valid for many other years as well: Surprise events, completely uncontemplated by forecasters, frequently dominate the outcome. The lesson to draw is that it is best to refrain from basing any important financial decision on a one-year forecast of the high-yield return.
No one would take seriously a one-day or one-week forecast of the high-yield return. Is it self-evident that one year is a time frame subject to a high level of predictability? Little if anything in finance is self-evident, but here are some useful facts that emerge from empirical investigation: In the 29 years spanning from the ICE BofA U.S. High Yield Index's first full year, 1987, through 2015, the yield-to-maturity (note 2) at the beginning of the year exceeded the annualized return over the succeeding five years in 26 out of 29 cases, or 89.7% of the time. In the three exceptions, the maximum amount by which the five-year annualized return exceeded the beginning yield to maturity, or YTM, was 0.37 percentage point (2001-2005).
Based on these findings, we can make a statement about the future that is genuinely useful, rather than akin to trying to predict who will buy the winning raffle ticket. It answers the question, "What five-year annualized return should we assume for a diversified high-yield portfolio that is being used, possibly as part of a broader investment portfolio, to fund five-year liabilities?" The most optimistic assumption one can defend is 5.35%, i.e., the ICE BofA U.S. High Yield Index's Dec. 31, 2020, YTM of 4.98% plus 0.37 percentage point.
Based on average results for 1987-2015, the best-supported assumption about the forward-five-year annualized return is 2.73%, i.e., the Dec. 31, 2020, YTM less 2.25 percentage points. That is because the historical mean difference between beginning YTM and five-year annualized return is -2.25 percentage points. Earning 2.73% a year for the next half-decade is an unexciting prospect. It is not an impossibly low figure, however, even though zealous high-yield advocates may reject it out of hand. Annualized returns were even lower in 1998-2002 at 0.52% and in 2004-2008 at 0.86%.
The reason that five-year performance is inherently more predictable than one year is that the single factor of beginning YTM explains fully 66% of the variance in five-year annualized returns versus just 26% for one-year returns. (R2 = 0.66 and 0.26, respectively.) This means that 74% of what determines the outcome over one year is determined by things that are unknown at the time of the forecast, e.g., the 12-months-later state of the economy, the level of underlying Treasury yields and the average spread-versus-Treasurys. (This is not to mention the unknown unknowns.) By contrast, only 34% of the five-year outcome depends on things not already known at the time that the forecast is being made.
On average, the one-year return differs from the beginning YTM by 8.05 percentage points, higher or lower. The comparable figure for the five-year annualized return is just 2.31 percentage points. By focusing on that narrower range, the five-year forecaster is likely to wind up closer to the mark than the one-year forecaster.
A forecast of the five-year annualized high-yield return is by no means guaranteed to score a bull's-eye, but in the context of the uncertainty in which financial professionals routinely operate, it is a prediction with valid applications in the funding of liabilities. Not so with a forecast over a horizon as short as 12 months. Investors hoping to generate alpha through short-run asset rotation strategies are best advised to ignore total return forecasts and focus instead on technical methods, such as momentum trading.
DECEMBER 2020 RECAP
Longest HY bonds performed best by far
The longest-dated (15+ years) high-yield bonds more than doubled the returns of high-yield bonds with maturities of 1 to 7 years during December 2020. This was despite highly negative returns on underlying Treasurys maturing in 10 years or later. Spread-narrowing of 34 basis points on the 15+ HY basket more than offset yield increases of 2 bps on the 10-year and 8 bps on the 30-year Treasury. The Treasury curve steepened during the month, as yields fell by 3 bps on the two-year and by 2 bps on the three-year, while being unchanged on the five-year and rising on the 10- and 30-year.
CCC & Lower issues climbed to positive territory for full year
CCC & Lower issues, which remained in the red through November, succeeded in posting a full-year positive return on the strength of a 4.11% total return in December. Even so, the bottom-tier issues turned in the lowest 2020 return among the three speculative-grade rating categories.
Leading the pack for the full year was the BB category, which combined income of 5.51% with a 3.11% price gain to achieve an 8.62% total return. Both the B and the CCC & Lower categories, on the other hand, showed price changes, measuring -2.83% and -5.32%, respectively. This all netted out to a full-year -0.02% price change for the all-ratings ICE BofA U.S. High Yield Index, which delivered a 6.17% total return.
Energy repeated as #1 in December, but finished last for the year
Energy far outpaced all other major high-yield industries in December 2020 with a 4.97% total return. That was despite finishing dead last for the year as a whole with a -6.62% return. Buoying Energy's December performance was a rise in the Generic 1st Crude Oil, West Texas Intermediate contract to $48.52, from $45.34. Aerospace, 2020's second-worst performer, at 0.45%, ranked second-best in December, at 2.94%.
Pulling up the rear in December were two classically defensive industries, Containers and Cable & Satellite TV. Neither suffered a fundamental setback. Both industries achieved price gains for the month to post total returns that would look quite respectable in a month in which the overall lCE BofA U.S. High Yield Index total return was less than December's outstanding 1.91% performance.
Trifecta returned 6.14%% for the month
Any short-term trader who correctly predicted December's top return categories by rating, major industry, and maturity basket, then put money on the combo, beat the ICE BofA U.S. High Yield Index's 1.91% total return by a healthy margin. During December, 41 issues of 14 Energy issuers had Composite Ratings in the range of BB and maturities of 15 years or longer. An equally weighted portfolio of those 41 bonds returned 6.14% for the month. Based on the choices made by Excel's "Remove Duplicates" function, an equally weighted portfolio of one bond per issuer — a more likely holdings list for a trifecta player — returned an even better 7.63%. While many, if not most, high-yield investors are longer-term-oriented than this analysis contemplates, the trifecta results highlight the profound impact of top-down decisions on interim returns.
Metals & Mining now cheap to ratings despite positive ratings prospects
The credit outlook is on the upswing, although it still has a long way to go to get back to neutral. In 2020, COVID-19 took its toll on companies' ability to generate cash flow. By September, not a single industry among the high-yield market's 20 largest had net positive ratings prospects (comprising both watchlists and outlooks). In November 2020, the positive-prospects count rose to one industry: Healthcare. Two more moved to the right side of the chart below in December: Cable & Satellite TV and Metals & Mining. In addition, Metals & Mining entered the previously empty northeast quadrant, indicating that it is cheap versus its peers on a rating-for-rating basis, even though the agencies indicate that its ratings are likely to rise.
The least attractive, southwestern quadrant (expensive despite negative ratings prospects) continues to contain such classically defensive industries as Consumer Products, Containers, and Food, Beverage & Tobacco. Investors willingly accept inferior rating-for-rating spreads on these industries, despite agency warnings that their ratings are likely to decline. Note that although Aerospace and Automotive & Auto Parts skew heavily toward negative ratings prospects, the market largely reflects that unfavorable characteristic, as indicated by those industries' location close to the diagonal line on the graph.
These conclusions are based on the methodology introduced in "New industry analysis shows defensives too tight." See the table below for each industry's coordinates. According to our analysis, industries located above (below) the diagonal regression line in the preceding chart are cheap (expensive) versus their peers, taking into account both their spreads, adjusted for ratings mix and positive or negative biases in their ratings outlooks and watchlists.
Short-term traders should switch to overweight on CCC & Lower issues
Short-term traders should switch from neutral to overweight on CCC & Lower issues. Our recommendation is based on the analysis presented in "When to over- or under-weight CCC & Lower issues." In December, the ICE BofAML CCC & Lower U.S. High Yield Index outperformed the ICE BofA BB-B U.S. High Yield Index for the second time after a period (October) of producing an inferior return. Under such circumstances, our analysis has found a statistically significant tendency of CCC & Lower issues to repeat their outperformance in the following month.
As always, we caution that this methodology is expected to deliver an edge over time rather than a win every single time that a nonneutral recommendation is generated. Note as well that at any particular time, the appropriate strategy for long-term, value-oriented investors may differ from the recommendation produced by this shorter-term, momentum-based approach.
Value investors should remain neutral on CCC & Lower issues
We are maintaining our neutral recommendation on the CCC & Lower sector for patient, value-oriented investors. As documented in "CCC & Lower relatively cheap in rich HY market," variance in the ICE BofA CCC & Lower U.S. High Yield Index's option-adjusted spread, or OAS, is mostly a function of the ICE BofAML BB-B U.S. High Yield Index's OAS. We calculate fair value for the CCC & Lower segment using the following regression formula, with spreads denominated in basis points:
2.34 x BB+B OAS + 73.44 = Fair value of CCC & Lower
In December, the BB+B OAS narrowed substantially to +328 bps, from +366 bps in November. Plugging the Dec. 31 BB+B OAS into our formula produces a fair value of +841 bps for the CCC & Lower index. By this measure, CCC & Lower issues were modestly tighter than their exact fair value on Dec. 31, with an actual OAS of +803 bps, down from +922 bps a month earlier. The minus 38 bps differential between fair value and actual spread compares with only -8 bps a month earlier but remains far away from our one-standard-deviation threshold of -254 bps for declaring an extreme overvaluation. Accordingly, we continue to recommend a neutral weighting for the CCC & Lower sector for longer-horizon, value-oriented investors.
US distressed should remain underweighted
As detailed in "How to tell when distressed bonds are attractive," the U.S. distress ratio, defined as the percentage of issues in the ICE BofAML U.S. High Yield Index quoted at distressed levels (OAS of 1,000 bps or more), can be used to derive the market's implicit forecast of the speculative-grade default rate. The distressed default rate — conceptually, the percentage of distressed issuers that default within one year — declines as the distress ratio increases. Multiplying the distress ratio by the distressed default rate gives us the market's expected 12-month default rate.
On Dec. 31, the distress ratio was 5.52%, down materially from an already low 7.29% one month earlier. The latest ratio is well below the 1997-2019 monthly mean of 13.82. Using the updated methodology described in "Estimating the market-implied default rate," we calculate an expected distressed default rate of 33.8%, indicating that the market is priced for a default rate of just 1.9% over the next 12 months. That compares with an implied Moody's forecast of 4.9%. We derive this estimate by multiplying Moody's U.S. bonds-plus-loans forecast of 7.7% by the 0.64 ratio of the agency's global bonds-only to global bonds-plus-loans forecasts.
At a gap of 1 percentage point or more, we conclude, based on historical analysis, that the distressed sector is likely to produce below-average returns over the coming 12 months. The present disparity of 3 percentage points far exceeds that threshold. Accordingly, we continue from last month our underweight recommendation for distressed debt within a high-yield portfolio.
Underweighting now recommended for high-yield versus investment grade
We are changing our previous neutral recommendation to underweight on high-yield in a portfolio that also contains investment-grade corporates.
In December, the option-adjusted spread on the ICE BofA U.S. High Yield Index tightened by 47 bps, while the investment-grade ICE BofA U.S. Corporate Index's OAS tightened by just 9 basis points. The difference in spreads, therefore, narrowed to +238 bps versus +321 bps in October. That reduced the spread to less than the threshold of +265 bps, below which we recommend underweighting high-yield. This conclusion is derived from the methodology introduced in "High-yield vs. investment-grade workout periods."
Underweighting still recommended on European distressed debt
As detailed in "European distressed debt fairly valued at present," the European distress ratio, defined as the percentage of issues in the ICE BofA Euro High Yield Index quoted at distressed levels (OAS of 1,000 bps or more), can be used to derive a leading indicator of the speculative-grade default rate. The distressed default rate — conceptually, the percentage of distressed issuers that default within one year — decreases as the distress ratio increases. Multiplying the distress ratio by the distressed default rate yields the market's expected 12-month default rate.
At the end of December, the European distress ratio was 3.97%, down slightly from 4.01% one month earlier. Using the methodology described in the above-referenced Feb. 24, 2016, piece, we calculate an expected distressed default rate of 26.0%, indicating that the market is priced for a default rate of 1.0% over the next 12 months. That compares with an implied Moody's base-case forecast of 2.4%, down from 2.8% one month earlier. We derive this estimate by multiplying Moody's European bonds-plus-loans forecast of 3.8% by the 0.64 ratio of the agency's global bonds-only to its bonds-plus-loans forecasts. The market's forecast undershoots the inferred Moody's base case forecast by 1.4 percentage points. When the European market-implied default rate is 1 percentage point or more below the Moody's forecast, we infer from our U.S. research that the distressed sector is likely to produce below-average returns over the coming 12 months. Because the difference between the Moody's forecast and the market-implied forecast exceeds that cutoff, we are maintaining our previous underweight recommendation on European distressed debt.
Global high-yield investors should remain neutral on Europe
During December, the European Equalized Ratings Mix, or ERM, spread tightened by 11.5 bps, while its U.S. counterpart tightened by 29.3 bps. The differential between the two regions consequently increased to +16.5 bps, from -1.4 bps. That change left the European high-yield in the moderately rich zone versus the U.S.
We recommend underweighting Europe only when it is extremely rich versus the U.S. Accordingly, we are maintaining our neutral recommendation on Europe within a cross-border high-yield portfolio. Europe is fairly valued when the ratings-equalized OAS differential between the ICE BofA Euro Non-Financial Constrained Index and the ICE BofA U.S. Non-Financial Constrained Index is in the range of 20.1 to 48.1 bps.
Our current conclusion regarding valuation derives from the ERM methodology described in "Europe-versus-U.S. valuation." The 2013 report introduced a technique for addressing the analytical challenge that Europe's high-yield market is much more heavily concentrated than its U.S. counterpart in BB issues. A direct comparison of quality spreads between the two regions mainly demonstrates that Europe's high-yield universe is less risky. It does not tell us whether investors are compensated more generously or less generously for a given level of credit risk in Europe than in the U.S. To remedy this problem, we equalize for the vast difference in ratings mix between the two regions to determine which region is trading cheaper, rating for rating.
Neutral weighting still recommended for emerging markets HY
We are continuing our neutral recommendation on emerging markets high-yield debt within a portfolio that also includes U.S. high-yield debt. On an ERM basis, the ICE BofA U.S. Emerging Markets Corporate Plus Index's ERM-based OAS tightened by 55 bps in December. Over the same period, the ICE BofA U.S. High Yield Index's ERM-based OAS tightened by 32 bps. The ERM-basis EM-minus-U.S. high-yield spread consequently ended December at +158 bps, down from +182 bps in November. That change in ERM spread just barely dropped the EM-minus-U.S. differential deeper into the third quartile of historical spreads. Because the third quartile indicates fair EM valuation, we are maintaining our neutral recommendation on emerging market high-yield debt.
The backstory on this recommendation is provided in "High-yield relative value, emerging markets versus U.S." That report introduced an analysis of relative value for emerging market corporates and U.S. high-yield. The key to this methodology is adjusting for the substantial difference in ratings mix between the ICE BofA High Yield U.S. Emerging Markets Corporate Plus Index and the ICE BofA U.S. High Yield Index. After that adjustment, the emerging markets sector invariably has a wider OAS than U.S. high-yield, but the magnitude of the differential varies widely over time. In the equity market, similarly, emerging markets are perennially accorded lower multiples than the U.S., reflecting concerns about the rule of law and corporate governance.
1. It is fair to ask me, a former sell-side high-yield strategist, whether I did not, in fact, participate in my firm's annual "Year-Ahead Forecast" ritual. That issue came up only at the last investment bank at which I was employed. Fortunately, my high-yield research unit was independent of the fixed-income research group, so for the first several years I simply chose not to participate. After a while, I realized there was an intellectually valid way to help out the buy-side high-yield research people who were compelled to fill in a year-ahead expected return box as part of their own firms' annual rituals. Each December thereafter, I published a total return projection, couched very carefully as the outcome that would be expected if economists' consensus forecast proved correct. In that case, the rating-agency base case default rate forecast probably would also be fairly accurate. I modeled the expected recovery rate on its documented relationship with the default rate. Full transparency about my methodology enabled buy-siders to be able to adjust the analysis according to their own in-house economic forecasts. My approach did not depend on my own firm's economic forecasts, nor did I endorse the consensus, which does not have a very good record for accuracy.
2. ICE Indices LLC reports yield-to-worst and effective yield data for the ICE BofA U.S. High Yield Index only from Dec. 31, 1996, onward.