91����

Because we work so closely with numerous financial organisations…many PhD students, are offered high-calibre jobs with them.

“For many years, our research has tried to address problems that banks and other types of lenders face,” says Professor Crook. “This ‘intensity model’ has evolved from earlier programmes of study that stretch back to the 1990s.” Traditionally, when a customer applied for a credit card, banks used data gathered from their application to predict the likelihood of defaulting – missing three payments – within a fixed time frame. This was revolutionised in the 1990s with the ‘survivor model’, which predicts not only if a person will default, but in which particular month it is likely to take place. The new generation intensity model devised by Professor Crook would not only predict if and when a borrower will default, but could also predict the likelihood of any individual account falling behind just once or twice, and the ability of making catch-up payments, on any given month.

The intensity model is not a new phenomenon. It has been used by statisticians as a means of explaining past financial activities, rather than predicting the future of individual accounts. Professor Crook is the first to use the model in this way. “We are going under the bonnet of credit card banking here,” explains Professor Crook. “And right now we are working to introduce macro-economic variables, such as interest rates, house prices and unemployment rates into the model.

“This may allow banks to more accurately stress-test their portfolio to ensure they are retaining enough capital to protect depositors, ultimately reducing the risk of getting into difficulties in the event of a future financial crash.” A further benefit will come into play from 1 January 2018, when a new international accounting standard requires banks to set provisions aside for every loan made in case it doesn’t repay.

“The provision is usually included in the interest rate,” Professor Crook explains. “However, if the risk increases after the loan has been made, intensity models can reassess what will happen during the life of the loan far more accurately than models being used today.”

The intensity model is good news for borrowers too: “If the risk of you missing a payment is more accurately assessed, your credit limit and interest rate will more truthfully reflect that risk. Someone that has a very high probability of missing a payment would typically be charged a higher interest rate, and people with low risk will be charged a lower one.”

Professor Crook developed the idea of using the intensity model to make predictions with Dr Mindy Leow, a former post-doctoral fellow who has now taken a position with a leading bank. Professor Crook says this is not unusual. “Because we work so closely with numerous financial organisations – sharing our ideas and providing consultancy support– many of the brightest minds, including a number of my PhD students, are offered high-calibre jobs with them in order to develop their concepts more competitively.”

As part of his role as Director of the University’s Credit Research Centre, Professor Crook and colleagues organise a bi-annual Credit Scoring and Credit Control Conference. Now in its 26th year, the event drew more than 400 delegates from 39 countries to Edinburgh in 2015. “The conference is the only one of its kind in the world,” he says. “Most industry conferences are run by a specific company, talking about their own developments.

Ours is open for anyone to present new ideas. This attracts all sorts of industry leaders – regulators, statisticians, credit bureau professionals and academics – and stimulates many lively and useful debates.” After the application of the statistical theory was initially made public in the European Journal of Operational Research in 2014, Professor Crook presented an enhanced version of the model to delegates at the Edinburgh conference in August. Reaction has been incredibly positive, although not overtly so.

“In private conversations, banks and other lenders have been very impressed,” Professor Crook explains. “But due to the extreme secrecy around credit risk models, and the need to ensure competitors are not aware of each other’s plans, we can’t reveal anything further on the matter at present.” While the intensity model heralds a better, fairer mode of lending and borrowing money, in reality banks will have to make millions of pounds of investment to implement the findings. There are also organisational hurdles to overcome before a model such as this one can advance. “Changing something so fundamental to a bank needs approval at many layers. And as the stakes are so high, it will take time for decisions to be made,” says Professor Crook. “Our model offers huge opportunities, and if lenders are willing to be courageous and imaginative, the UK could pave the way for more effective and accurate credit risk assessment.”