One of the aspects of the CEBS European stress test exercise that has been commented upon quite widely is their decision to only apply haircuts to sovereign debt held on the trading books of the banks examined. However, most of these bonds are held on the “banking books” on the understanding that they are being held to maturity and the CEBS exercise assumed no sovereign defaults over the time horizon considered, so no haircut was applied to this portion of the bond portfolio.
In reality, the trading book\banking book distinction is arbitrary. In the case of a default or restructuring on these sovereign bonds, the distinction is meaningless. In the case of a bank failing, the distinction also doesn’t mean much: If the assets of the bank need to be sold off to meet liabilities, then bonds originally intended marked as hold to maturity the banking book may still have to sold off at market values.
I’ve come across two interesting alternatives to the CEBS stress tests. The first is this report from the OECD, which takes a macro look at the topic. They calculate that 83% of the exposure to EU sovereigns is held on the banking book and the report gives a good sense of the exposures of banks in different countries to various types of sovereign risk.
The other alternative comes via the Calculated Risk “Some Investor Guy” series on sovereign debt. The Guy linked to this rapid response piece from Citi, redoing the analysis on a bank by bank basis. Applying the haircut to the banking book as well as the trading book, the number of European banks that fail the test rising from 7 out of 91 to 24 of 91.