Professor Patrick Honohan:

There are certainly big differences. First, I am not a great enthusiast for stress tests because they are very partial. There certainly have been changes in the technology. Stress tests nowadays are accompanied or preceded by a detailed file-by-file sampling and analysis of whether a given bank really has the declared loans, whether they are as good as they look and whether the summary management information that the stress tests are based on are soundly based on actual loans that are properly collateralised. This is the basis of an asset quality review, AQR. It is very costly. We have spent tens of millions of euro on asset quality reviews during the period of the bailout and heading into the single supervisory mechanism. The idea is that we know that the basic data is okay and revisions have been needed for that.

The stresses used in 2006 were not very strong. That is evident to everyone. One test was whether a bank would survive if there was a 5% fall in GDP next year and a given fall in house prices. The stresses were not severe enough. We can see the stresses now. They are quite severe in the latest test. In a risk environment, especially with the books that the banks have - they are still exposed with many non-performing loans - there is always uncertainty.

There is a tendency for people to exaggerate. For example, in 2011 we did a very big, hard and expensive job for the stress test. It was much more elaborate than 2010. We can talk about 2010, but that will be for another day. However, we were not sure. Some people said, "Nothing can possibly happen from now on." However, it was a great relief, when we did the exercise in 2013-14, that we were okay.