V Shaped Price Recovery
The largest part of the outperformance is also perhaps the simplest to explain. “Fallen angels become technically oversold during the 6 months running up to their downgrade date, but then recover most of the loss over time,” explains Francis Rodilosso, Head of Fixed Income ETF Management at VanEck, who has featured in various fixed income podcasts and authored blog postings on the VanEck website.
“They on average drop 8-9 points and recover 90% of this during their first bounce, which occurs somewhat symmetrically within about 6 months of the downgrade,” he continues. The reason for this is forced selling from ratings constrained investors – those tracking, or benchmarked against, an investment grade bond index. And since the IG asset class is currently almost 6 times larger than the HY asset class, it is a plausible hypothesis to suggest that the volume of selling by IG investors outweighs the volume of buying by HY investors. (Some active credit investors do specialise in “crossover” credits, which include some fallen angels soon before and after downgrades, but their activity does not seem to have arbitraged away the fallen angels anomaly).
The next leg of fallen angels outperformance comes from longer term price appreciation. Most fallen angels enter the index at a low 90s price, and some of them are eventually redeemed at par. Many bonds however will exit the Fallen Angels index upon an upgrade, and reascend to investment grade, before they reach maturity. “Over 40% of them reascend to investment grade typically over a timeframe of three years”.
Top down sector contrarian bets
Beyond technical, index and rating driven trading, another prism for explaining this longer term outperformance has been serendipitous and is more “top down” in nature. Fallen angel indices tend to end up making a hefty contrarian wager on out of favour sectors, which have included autos, banking, telecoms, and energy, all of which have subsequently recovered in the past.
“Bank paper became a big part of the index after 2008, at one stage reaching 25% of the US index,” recalls Rodilosso. ( The ICE indices decided not to include CoCo or Contingent Convertible bonds).
Around the very start of the index in 2004, autos were important (and reaching back earlier to construct a synthetic index, telecoms would have been important).
Energy also became a larger weighting after 2015 (The sector contrarian effect is not easy to test statistically as there are not enough observations).
Higher Credit Ratings than High Yield
Fallen angels have a higher average credit rating than high yield, with most fallen angels clustered around the BB rating, which provides some degree of ballast. “During many periods of spread widening and decompression, BB bonds tend to outperform B and CCC rated bonds, points out Rodilosso.
“Although sometimes the rating effect is outweighed by the higher duration of fallen angels,” he admits.
Reasons for years of underperformance
Fallen angels nearly always have longer duration than high yield, averaging around one year longer, which is not surprising when original issued IG bonds are longer term than HY bonds. (In 2008, fallen angels briefly had shorter duration than high yield for technical reasons related to bond mathematics of modified duration. “Fallen angels usually have a lower average dollar price than the broad universe, but that gap became even larger with a 20-30 point gap. Fallen angels did however outperform in both 2008 and 2009,” he explains.
In 2022, VanEck’s US Fallen Angels strategy had 3 years longer duration than investment grade indices, and did somewhat underperform high yield, while its Global Fallen Angels strategy somewhat outperformed.
Duration can sometimes explain underperformance, but equally there have been periods of rising rates when fallen angels outperformed high yield. “In 2009 and 2014, both the US and global versions outperformed despite five year yields moving up by over 100 basis points,” recalls Rodilosso.
“In 2018, fallen angels also lagged high yield, in part due to rate rises but also thanks to the sector mix: it had more energy than broad high yield. In 2007 the underperformance was more to do with the sector mix, and in 2011 it related more to quality,” points out Rodilosso. Fallen angels can end up with a higher concentration in certain sectors than the larger IG or HY universes
Simple indices are easy to replicate
VanEck‘s strategy is almost entirely passive, bar a small amount of discretion around the selection of issues and the timing of execution.
This makes index selection paramount. VanEck has selected the oldest continuous fallen angels indices, managed by ICE: the ICE US Fallen Angel High Yield 10% Constrained Index, and the ICE Global Fallen Angel High Yield 10% Constrained Index.
Conceptually, it defines a fallen angel bond by the instrument and not the company credit rating. This is an important nuance. “Sprint, which exited the FA index last year, had been in it since 2008 and 2009, but the ICE index would only include those bonds originally issued as IG, and not those subsequently issued as high yield between 2012 and 2015,” explains Rodilosso.
It also has a simple structure that is relatively easy to replicate. The VanEck ETFs own nearly all names in the ICE indices. The US Fallen Angel ETF owns all 72 US issuers, and the global ETF owns 145 out of 152 globally. Annualised tracking error is below 1% and was only 0.55% in 2023.
“It also has sensible rules on liquidity guidelines around minimum issue sizes,” he says. (“Reasons for omitting a small number of issues could include them simply being too close to other bonds on the curve, lacking liquidity, or having too high a minimum trading size, possibly due to them being Reg S,” elaborates Rodilosso).
VanEck have never needed to invest in anything outside the indices nor use any index product (in any case no standardised index CDS exists for Fallen Angels).
The main diversification constraint is 10% per issuer, which happens to overlap with one part of the UCITS 5/10/40 diversification criteria that applies to the European version of the ETFs; VanEck has never needed to downsize a position to stay inside the UCITS rules.
Beyond selling bonds with under one year to maturity, there are no time constraints. “Some trading strategies would overweight the first 6 months of a fallen angel’s life because most of the price recovery has historically occurred in that period. However, that would also make the index less diversified. The universe is already quite small, there are only 72 US Fallen Angel issuers and 152 in the global universe,” points out Rodilosso.
The index has no sector constraints.
Execution and trading
VanEck’s fallen angel ETFs have been 100% invested in index names, and use cash bonds rather than CDS. “We have always been able to do custom trades with ETF market makers and find there is plenty of liquidity in bonds entering or exiting the index,” says Rodilosso.
He does not wish to provide too much detail on the exact timing of rebalancing trades around index additions and deletions, for fear of alerting the market: “though we have not noticed any “front running or gaming” activity we are vigilant nonetheless,”.
Rodilosso can however mention that the ETF is not forced to sell defaults immediately. “In rare cases of defaulted bonds, we can be somewhat tactical in locating distressed debt specialists who are keen buyers for these sorts of bonds”.
UCITS and SFDR
The UCITS version of the product does not employ security lending and has never needed to employ its leverage facility, which would anyway only be used to bridge small time gaps around settlement periods.
For the time being the two ETFs make disclosures under article 6 of SFDR. VanEck could consider moving to SFDR 8.
US vs Global exposure
There is some overlap between the global and US strategies. The global fund is c.35% comprised of the US fund names, c.28.5% in other developed markets, and c.36.5% in emerging markets (but only has 1.1% in China including Hong Kong).