Mr. John Beggs:

I provided the committee with a statement on this matter on 27 March. Well, first of all, I confirm that I did meet with Professor FitzGerald in relation to stress testing. I do not recall the exact date but I'm happy to accept from his records that it was in October 2005. Professor John FitzGerald himself, in his e-mail of 16 February to the committee, accepted the possibility that his recollection of the content of the meeting was faulty. Professor FitzGerald's recollections of the content and the nature of the meeting were in many key respects very different from mine. Specifically, I am certain that I expressed no concerns on the part of the AIB board about stress testing. The board never authorised or was aware of the meeting. I did not report back to the board on the outcome of the meeting. The meeting was an exploratory, technical one between economists to ascertain whether the ESRI's model could be used to enhance the scene setting of base and stressed economic scenarios by providing more economic variables than contained in regulatory exercises at the time. It was never the intention to ask the ESRI to carry out stress tests independently of regulatory ... of the regulatory ones and it was certainly never envisaged that the ESRI would apply macroeconomic scenarios to the bank's internal data. I never asked Professor FitzGerald to carry out stress testing on behalf of AIB. I sought to explore whether prescribed regulatory economic stress data, which consisted of only a limited number of indicators, could be run through the model to generate a more graphic and detailed economic picture of the stressed outcomes. The meeting failed to produce any results. As I recollect, Professor FitzGerald indicated that the ESRI would not undertake private unpublished work. There were some limitations as to what the model could do, but he was prepared to make data from the forthcoming medium-term review available to us. It was an inconclusive meeting and my unit never followed up on that.

If Professor FitzGerald did any other work for AIB or other banks on stress testing, I am not aware of it.The impression created by Professor FitzGerald's testimony is that the AIB board was concerned about the lack of severity in the stress test exercises as far back as late 2005, which may imply a broader level of concern. I have no knowledge of the board's opinion on the matter. I never received any negative feedback or reports from the board or board risk committee on stress testing. As a result, I have no reason to believe that the AIB board had any misgivings about regulatory stress testing. It was not my practice in AIB to attribute comments or opinions to other AIB business units or committees, least of all the board of the bank.

Stress testing in AIB is executed through the stress testing steering group, which is responsible for ensuring that a comprehensive stress testing programme is embedded in risk management and to ensure that an effective framework is in place to enable stress testing across the group. I attended the meetings of the stress testing steering group when a stress test exercise had to be carried out. My role was to communicate the evolution of the scenarios from base case to stress scenarios to the committee and onward to divisional risk management and business units, as appropriate. I was not involved in the application of the economic variables to the bank's internal data. This was ... sorry ... carried out by analysts within the divisional risk and credit units as well as in the group risk unit.

From an early stage, probably prior to 2005, I felt that the regulatory stress testing exercise could become too procedural. Banks were provided with stress tests, the work was carried out with great care and attention to detail, signed off by bank boards and reported on by the Central Bank in various reports, such as the annual financial report. I also felt that the stresses were too mild, though not on every occasion. I was also concerned that the moderate nature of the stresses were not conducive to maximising management buy-in of the risks involved. This was my personal opinion and had much to do with the view that people performing the stress tests in various business units needed to understand the broader economic and financial implications of a given shock to a base case scenario. To do this, one needed a model and the ESRI had one. A model would provide more outputs for consideration, a wider context to the changed economic environment and given the likelihood of a growing stress testing framework, a more consistent approach to the exercise over time. I raised these issues with the stress testing steering group and suggested that I talk to John FitzGerald in the ESRI to see what could be done. I stated this would have some cost implications.

I quote John FitzGerald from his testimony to the committee:

We did macro-economic scenarios. The difference compared to what the Central Bank and so on was doing was that we used a model, so it was consistent. One did not get a housing price crash and no change in unemployment; one got a housing price crash, unemployment going through the roof and Government revenue collapsing, all coming together which is the way to do it. That was my concern. It was not just the Central Bank. It was interesting doing something similar with a British owned bank. The Bank of England and the Financial Services Authority, FSA, had a similar inappropriate approach to stress testing.

So, I also felt that the amount of data provided and scenario scene setting between the base case and stress tests were too limited. This applied to both Irish and UK regulatory tests. I wanted to bring more information to bear on the process of assessing the risks to the business from the economic deterioration under consideration in the stress. I carried out, personally carried out, most of the work on the macroeconomic inputs for stress testing in AIB. From year to year, I endeavoured to be consistent in how I interpreted the shocks and in producing additional information for the divisional risk and business teams, particularly the changes in employment, which were not specified in the macro-aggregates in the base or stress scenarios. This had to be estimated within individual banks from the changes in real GDP and unemployment rates supplied for the exercise.

In his testimony, Professor FitzGerald stated on several occasions that his recollection of the meeting in October 2005 might be faulty. In his letter to the committee of 16 February he states: "However, as my recollection of the timing of the meeting was faulty, there is always the possibility that my recollection of the content of the meeting was also faulty." Mr. Chairman, much emphasis has been placed on the importance of this meeting between the ESRI and AIB in October 2005 at the committee hearing on 11 February but it did not have the imprimatur of the AIB board. Thank you.