The newly released stress test of selected EU-area banks by the Committee of European Bank Supervisors (CEBS) is flawed in its methodology and the results are not a reliable indicator of EU bank sector soundness. A stress test should examine the impact on net portfolio value of extreme but plausible shocks to the key variates explaining net portfolio value. The CEBS report states proudly and repeatedly that it uses extreme but plausible shocks, and this is true, but it ignores the key-variates criterion of a well-designed stress test.
We know that historically some of the key drivers of systemwide bank failures are falls in asset prices (particularly property prices), sovereign debt defaults, and liquidity freezes. Sovereign debt defaults and liquidity freezes are explicitly excluded in the CEBS test. Falls in asset prices are included only to the extent that they are generated by the two shocks used in the test: a fall in GDP growth rates and an increase in bond market credit spreads. This specification gives rise to a very relaxed stress scenario.
Consider for example the case of Ireland. The baseline one-year property price change for Ireland in the CEBS model is minus 13 percent. This is reasonable as a baseline case although it seems a bit optimistic. What is not reasonable is the supposedly extreme-but-plausible adverse scenaria – a one-year decline of minus 17 percent! With most independent analysts predicting continued big falls (e.g., Morgan Kelly, Kevin O’Rourke, Ronan Lyons) this should be substantially larger. The extreme but plausible scenario should have a fall of minus 50 to minus 60 percent over the next two years; minus 17 percent in the first year is not extreme enough for a stress test.
Similarly, in the cases of other vulnerable markets which have experienced big property price bubbles since the advent of the Euro, the report’s adverse scenario relies on much too mild property price declines. For example, the extreme-but-plausible adverse scenario for Spain uses a property price decline of only minus 5 percent.
The CEBS report explicitly excludes the possibility of a Greek sovereign debt default. The argument made in the report is that such an event is rendered implausible by the creation of the EU bailout funds, in particular the 110 billion joint EU-IMF fund for Greece and the 440 billion European Financial Stability Fund available to all members. Yet Ken Rogoff was quoted only one week ago (PBS Newshour 19 July 2010) stating that he believes a Greek default is not merely likely but, over the medium horizon, is virtually inevitable: direct quote: “They are going to restructure their debt, they are going to default.” Rogoff’s views are widely respected around the world and he is not inclined toward exaggeration. How can the CEBS describe as implausible a view that is so widely held by so many respected analysts? Does not this view at least qualify as extreme-but-plausible?
There are two problems with the CEBS report’s claim that the bailout funds have rendered a Greek sovereign default implausible. One, the bailout funds are supposed to provide conditional support, premised on Greece (or other troubled sovereigns) continuing with the painful austerity measures laid out in the bailout fund agreements. Does the CEBS believe that the conditionality in the bailout fund agreements is not credible, and that the bailout funds will be accessed in times of panic now matter how the defaulting government behaves? If so, this shows a refreshing lack of faith by one EU institution for the promises made by another.
The second problem with the bailout funds is that they are not extensive or long-lasting enough to permanently eliminate the risk of default. These expensive big funds will serve to slow down a Euro-area sovereign default, but not to prevent one eventually. In the interim, the funds will be another big subsidy to the banking industry, and might serve as a juicy source of profits for hedge fund managers willing to bet against the EU political establishment.
Here is my suggestion for an alternative specification for an EU-area bank stress test. It uses a four-year rather than two-year horizon. It relies on two extreme but plausible shocks: a Greek sovereign debt market default/restructuring and country-specific average inflation-adjusted property prices falling to their 1999 levels. Note that the second shock requires bigger property price falls in countries like Spain and Ireland that experienced property price bubbles post-1999 and smaller falls for countries like Germany and France which did not have these bubbles.