CDS: Examining the negative basis

In our Fixed Income Chart of the Week for 8 June 2020, we noted that the CDS basis for names in the Markit iTraxx Europe index had gone significantly negative in March, which struck us as very unusual. Even though spreads for both CDS and cash bonds widened simultaneously, the greater liquidity of credit default swap indices meant that movements were less pronounced in that segment of the market and positions reversed more quickly. The discrepancy is unlikely to persist, however, as it opens up potential arbitrage opportunities, and the gap has already started to close again.

First, let us remind ourselves of two pieces of terminology:

  • Basis = CDS level minus the comparable point on the bond curve. In this blog post, we are focusing on the 5-year basis.
  • Skew = Index level minus theoretical level, obtained by averaging the CDS level of individual names.

For our analysis, we are using the CDS data from Markit, and Axioma Fixed Income Spread Curves.

The 5-year CDS basis is in significantly negative territory, which is unusual. Although we have focused on the European index, the same is true for the CDS investment-grade index in the US. As most of the world went into lockdown, both bond spreads and CDS levels widened. In a crisis that is expected to be short-lived, buying protection makes sense, and we believe this is what banks and hedge funds did. The easiest way to buy protection is to buy the index. Real-money funds would not necessarily have had this ability, and many would have needed to sell bonds. It was only afterwards that the basis widened. Following the initial panic, we believe some of the index hedges would have been closed out, but there has not been a rush back into corporate bonds. This difference in behaviour between two distinct groups of investors has left a gap between the CDS index and bond spreads.

The difference between a CDS index and bond spreads can be split into two parts: the movement of indices relative to single name CDS, as measured by the skew; and the difference between single name CDS and bonds, as measured by the basis. If we look at a chart of the skew, we can see that while it did exhibit a relatively large negative move, the minimum skew was -21bps, compared with -79bps for the basis.

The much smaller move in the skew we attribute to less liquid single name CDS being priced relative to the index, thereby limiting the extent to which the skew can move, and explaining why it is the basis that shows the majority of the move.

Repeating our analysis, this time splitting by rating, we see that similar directional moves are apparent in each ratings bucket, although the moves are exaggerated in lower-rated higher-beta names, as might be expected.

Interestingly, the negative basis was initially far more pronounced in financials than in non-financials, but this difference did not last.

Overall, the move in the basis is not driven by a single part of the market, but seems to be apparent across the whole of investment grade. It seems unlikely this will persist, as a large negative basis at an individual name level could lead to a resurgence in negative basis trades (buy the bond, buy protection, keep the carry), as were popular before the global financial crisis, themselves reducing the basis.

It will be interesting to see where we go from here.

Alan Langworthy

As part of the multi-asset research team, Alan's primary focus is on the next generation of Axioma's fixed income risk model. Prior to joining Axioma in 2016, Alan spent fifteen years at UBS, the last ten of which were in UBS Delta where he worked on risk, performance attribution, liquidity, and bespoke client projects.