Martin Wolf appends this chart to his most recent column. It shows the well-known-by-now interbank spread – the way that Libor has gapped out relative to central bank rates – and tries to decompose it into two parts: a credit premium and a non-credit premium. According to the chart, there’s a big difference between Libor spreads now and Libor spreads back in August: in August they were mainly made up of non-credit premia (illiquidity, that kind of thing), while today it’s all about credit risk.
I don’t buy it for a second. If you read the small print, it says that for the methodology underlying the chart, we should see the Bank of England’s Quarterly Bulletin for Q4 2007. It’s a very large PDF file, so I’ll simply tell you what it says:
The analysis in this box uses prices of credit default swaps
(CDS) for banks in the Libor panel to form a rough estimate of
the credit premia implicit in Libor rates.
In other words, if Libor spreads gap out, that’s not necessarily due to credit risk. But if bank CDS spreads gap out, that’s definitely credit risk.
Ahem. This is a world where the US Government has a CDS spread of 16bp; where agency CDS is something over 200bp; and where, more generally, there’s a huge number of forced unwinds going on in the CDS market. As I noted in a blog entry with the title "CDS: It’s Not About Credit", the FT itself is printing articles explaining that companies can raise new funds significantly inside their CDS spreads.
In general, it simply doesn’t make sense to use CDS as a proxy for credit risk, especially not when you’re dealing with credits as volatile as banks. Indeed, the fact that the non-credit premium on this chart actually dipped well below zero recently should prove that there’s something very wrong with the methodology. It’s a silly chart, and Martin Wolf shouldn’t have printed it.