I have previously discussed the topic of negative basis trades, specifically as manifested in cash-CDS pair trades. An interesting tangent on the basis trade topic, and one that is causing many index arb funds a lot of headaches lately, is the IG Index basis, or the basis between the IG11 (current on the run) index spread and the wighted average spreads of the Fair Values of its 125 constituents (listed here). If the index trades tighter to its Fair Value, the basis is negative, and an account can fully hedge a basket of equally-weighted IG constituents, while picking up 40 basis points.
Historically, the IG index basis has been positive, with a September 15 inflection point, when the basis flipped, reached a peak spread of 58 bps by the end of 2008, and is currently down to 40 bps.
The reason for the historical positive basis was a result of index funds being net sellers of the basis package. The blow up in basis, due to factors which I will present shortly, has made the existence of index arb
funds virtually impossible. Curiously, the negative basis collapsed rapidly in mid-January only to rip wider subsequently. The brief blip of rationality may explain why several large index arb funds posted
January returns; current spread data indicate poor index arb
performance will persist.
As the markets went haywire "after Lehman" the IG index spread collapsed, for the following reasons:
- Dispersion between specific single names and IG index
- Abnormal unwind costs and jump to default risk
- Upfront cash payment on index due to low index strike
- Illiquidity in certain single names
- Mod-R vs Ex-R restructuring language
- Single name technicals
1. Index-To-Single Name Dispersion
A long, long time ago in Q1 2007, IG8 had a fixed coupon of 35 bps and the widest spread on any of its underlying constituents was 185 (Rescap), with 85% of the portfolio reading inside 60 bps. The current on the run index IG11 trades at a 150 bps fixed coupon, with 62 out of 125 names trading much wider than index strike, and has three outliers that trade on a point up front basis (500 bps running), Textron, MBIA and iStar. The aberration from preponderance of wide "outliers" creates a duration mismatch between the Index and the FV legs.
As wide spread names have pronounced convexity, the trade equalization starts to lose practical sense once more and more names escape to the risk stratosphere. The index attempts to compensate for this by readjusting every 6 months, by kicking out outlier names, as the index moves to a new on the run iteration. One can be sure that IG 12 will not hold Textron, MBIA and iStar, as well as at least another 30 current index constituents.
2. Abnormal unwind costs and jump to default risk
As anyone who has ever tried to novate (or merely unwind) a deep in the money trade with a broker/dealer, that image of dollars in your head comes quickly crashing down once the counterparty gives an effective price at which it is willing to unwind the trade. The reason for this discount to market is that the B/D has to take on jump risk by paying you a significant upfront cash payment, and perform an opposing transaction in the market at a less opportune inception basis. As B/Ds need to make sure their counter-trade is viable until maturity (via collecting on the annuity) in order to offset the upfront payment, they will likely also buy additional short-dated protection, as well as charge a premium for the willingness to assist in unwinding...
A simple way to avoid getting the "Goldman discount" is to take on the jump to default risk yourself. Instead of novating a name, simply purchase the opposite CDS at the desired unwind price. The differential between the in-the-money strike price and the unwind cost creates an annuity whose PV would equal the cash payment from the B/D, however it would stop if and when the underlying reference name were to default.
3. Upfront cash payment on new index purchases
As noted, the IG has a fixed coupon of 150 bps, so any purchases at a wider strike price would result in a net initial cash outflow. Indicatively, purchasing 100 million IG11 at the current 200 bps price would cost $2 million dollars upfront net of accrued interest. Many smaller accounts are unwilling to post such large amounts of collateral apriori.
4. Illiquidity in single names
IG11 is relatively liquid, and while a far cry from 2007 when index bid/ask spreads were about 1 basis point, current bid/asks are in the 2-5 bps range, depending on what the VIX is doing. At the same time some of the wider single name constituent outliers have 25-50 bps bid-offer spreads. If one recomputes the Index basis using tradeable bid/offer levels instead of mid levels, the negative basis become even wider.
5. Associated restructuring language
IG11 trades with Ex-R restructuring language, while single names are usually traded Mod-R, which is also usually substantially more liquid. One can interpolate the Ex-R:Mod-R relationship, and the 40 bps basis is calculated using Ex-R Single name spreads, however, the very existence of this dichotomy could generate ongoing basis aberrations as clients are more focused on relevant restructuring contract language.
6. Single name technicals
Additional single name technical pressure has arisen from the need to manage jump risk (for both managers and broker/dealers) due to risk managers finally waking up, as well as the continued market disruptions from ongoing CLO unwinds.
The result is that index arbs, which used to game the positive basis actively in "Pre-Lehman", will likely be out of the market for a long time, and if any of them have not yet fully unwound existing positions, they will likely be in for a continuing world of pain.