Negative basis trading

The FT recently discussed negative basis trades. Here is the basic idea.

• A bank buys a bond, typically a long dated one.
  
• The bank buys a CDS or a financial guarantee policy to maturity of the bond from a counterparty, often a monoline.
  
• The bank would then hedge against the risk that the protection it had bought was ineffective often with another monoline.

This was profitable despite the multiple layers of protection since the credit spread of the bond was bigger than the cost of the first and second hedges combined. Remember a bond spread includes compensation for much more than the risk of default: it includes compensation for illiquidity, for the volatility of the value of the bond, and so on. The bank is basically monetising those premiums.

Most of these trades were done in the trading book so the banks concerned booked the PV of the difference between the credit spread of the bond and the cost of the protection up front.

With the monolines at AAA and monoline protection some tens of basis points, this approach was not a huge issue. But now monoline protection is hundreds of basis points and the AAA ratings might not be with us very long. Also, the bonds used were often either the supersenior tranches of CDOs or long dated inflation linked debt. The latter isn't a problem: the former is, since the underlying credit quality of these bonds has gone South for the winter too. And negative basis trades are out there in size. The FT says:

Bob McKee, an analyst at Independent Strategy, a London research house, believes that up to $150bn worth of CDO business done by the monolines could be negative basis trades.

Doubtless there are some who will use these trades as another stick to beat the already bloodied body of structured finance. I would suggest the reality is somewhat different. The problem isn't the trades themselves: it is the selective use of mark to market. Marking the trades up front is fine providing you do it properly. That means:

• Credit adjusting the pricing of all derivatives. The details are complex here but basically you PV the value of net cashflows from a counterparty back along their risky credit curve rather than along the Libor curve. This has the effect as a counterparties' credit quality decreases of automatically marking down your trades.
  
• In particular valuing trades with realistic default correlation assumptions. In particular the only time that you need written protection on supersenior ABS is when the ABS market is in trouble - and that is just when the monolines are in trouble. Therefore the probability of joint default of the bond and the protection seller is not
  
  PD(bond defaults) x PD(monoline defaults)
  
  As it would be if they were independent. Instead it is something a lot closer to
  
  min(PD(bond defaults), PD(monoline defaults))
  
  Since the default correlation is so high. For the full negative basis trade it is reasonably close to
  
  min(PD(bond defaults), PD(monoline1 defaults), PD(monoline2 defaults))
  
  The real problem, then, is that some banks may have used naive default correlation assumptions in marking these trades and hence they are carrying them at an inflated value.
  
• Using realistic funding assumptions in valuing the position. I shudder to think about this, but it would not be massively surprising to discover that some of these trades were also valued under the assumption that the bank could fund the bond at Libor flat forever. That means in effect that the position has again be overvalued up front and will show a net carry loss over time.

Of course none of these issues would have seen the light of day without the declining credit quality of the monolines. But it does highlight the fact that those banks which have prudent P/L recognition and state of the art valuation policies are much better placed to withstand market turmoil than those who don't.