Patrick Shallow
PhD Thesis Title: Empirical Measurement and Analysis of Retail Credit Risk.
Supervisor: Professor Louis Murray
External Examiner: Professor Philip Molyneaux, Bangor University
Abstract
The subject of this thesis was how, through the use of economics, mathematics and statistics, it might be possible to measure credit risk in retail lending.
The author, Patrick Shallow had found professionally in a bank to which he was appointed in 1980, that the bank’s bad debt provision requirement had proven inadequate to emerging arrears, which then led the bank into insolvency. He had noted that the credit risk matrix of the parent bank in the UK had been deployed in the vastly more stressed Irish economy. Using, initially rudimentary, regression analysis, he discovered that the UK risk matrix was manifestly inadequate to accelerating credit risk in Ireland. He then set about developing a model which would begin to measure actual credit risk in Ireland, as evident from the various monthly statistics, relating to growth, loan composition, loan maturity, portfolio seasoning , arrear migrations , plus write offs and final recoveries.
He noted in 2001, when he commenced his research, that under the Basel 1 reporting requirements, various arbitrary estimates were allowed to be made against the up to date/ well performing loans. Post 2006 he then found that, even worse, to counter the banking tendency to exaggerate risk in the well performing portfolio for tax purposes, no provision was allowed under the regulations against these loans (typically representing some 90 % of all loans )
In the course of his consulting work, Patrick was invited to work with one leading bank to measure its credit exposures. Equipped now with a data base of some €3billion in consumer finance loans he was able to develop a dynamic model which first segmented all loans into four disparate portfolios, viz. Large Secured Agricultural, Secured Consumer Finance, Unsecured Consumer Finance and Micro Finance. Then using rolling averages he was able to observe arrear trends. Because of the richness of relevant data, he was then able to analyse results from the macro to the micro level and to distribute portfolio –level credit risk across all accounts and to arrive at Portfolio Loss Forecasting, Loss Provisioning policies, profitability analysis ,credit risk pricing to eliminate cross subsidisation and additionally to facilitate individual loan loss forecasting and performance management.
The bank client adopted the model, now entitled Empirical Credit- Risk management (ECM), which it has since that time. The initial desk application product was developed into a cloud based version, was patented and has been deployed into a number of banks and credit unions, with the approval of the leading auditors and the central bank of Ireland. It is now being exported to foreign banks.