Professor Patrick Honohan:

What they did was, and I'm sure there are other firms that do that nowadays, so I'm not making an emotional ... they had a research group ... a consultancy group that ... what they brought was an insistence on collecting as much data as possible and confronting that data with little models that they had that projected what losses would occur depending on what you knew about those loans. So, you take a loan and you say, "What do we know about this loan?" If it's a residential mortgage, you mightn't know an awful lot - you'd know the location, you'd know the loan-to-value ratio at origination, you might know something about the income of the borrower at origination - these could be used to predict. Also, you will know ... because of the location, you will know what the unemployment rate and the trend of unemployment is in that location. So you use that sort of information to project forward and over ... actually, they projected over 40 or 45 years, the likely loan losses. To some extent, it's ... their models are over-ambitious, but they distance the calculation of losses from the emotion and the touch-and-feel of the loan officer and the attitude of the bank management. They put it ... distance it and put it into a mathematical box and it produces a result, which may not be correct, but it's not going to be a result that has been filtered by rose-tinted spectacles.