Basel II introduced a new and more sophisticated way of modelling banks credit risk which is used by the big four banks in Australia. A recent study uses advanced mathematical techniques to stress test this model.
Silvio Tarca CMCRC PhD Researcher U. Sydney Keywords: Banking, regulation, predictive modelling Silvio Tarca has experience in scientific research, investment management, information technology and engineering at top tier organisations including Los Alamos National Laboratory, J.P. Morgan Asset Management, Accenture and Timken.
A new study by Silvio Tarca and Professor Marek Rutkowski assesses the accuracy of the Basel II credit risk model by comparing its estimate of regulatory capital against computer simulated models and finds that the model is accurate. In 2008 the Australian Prudential Regulation Authority (APRA) implemented the Basel II Accord for determining the capital requirements of banks and other deposit- taking institutions that it regulates. Basel II requires that banks hold capital for credit, market and operational risk to absorb losses and provide protection against insolvency. APRA requires banks to determine their capital requirements for credit risk by applying either the standard method or the internal-ratings based (IRB) approach. The standard method categorises loans and other assets by asset class and credit rating grade and multiplies the credit exposure of each category by prescribed risk weights, ranging from 0% to 400%, to arrive at an estimate of risk-weighted assets. The minimum capital requirement is then simply calculated by using 8% of riskweighted assets. The standard method, which is an extension of Basel I, is straight forward for banks and other deposit taking institutions to administer and produces a relatively conservative estimate of regulatory capital. The IRB approach on the other hand calculates capital requirements directly using an asymptotic (a way of exploring real-world phenomena in mathematical modelling) single risk factor (ASRF) model. More sophisticated than the standard method, the ASRF model requires more data and its greater complexity makes it more expensive to administer. However, this model typically produces lower capital requirements than the standard method and as a consequence banks using the IRB approach may deploy more capital in pursuit of more profitable lending opportunities. Given this, it is not surprising that all four major banks in Australia use this ASRF model while the complexity of the ASRF model means virtually all other deposit taking institutions use the standard method. In his study Silvio tests the accuracy of the ASRF model on the balance sheet of a hypothetical bank constructed to be representative of the loan books of the four major Australian banks. Comparing regulatory capital calculated by the ASRF model with that computed by simulations he finds empirical evidence that the IRB approach is accurate. The Capital Markets Cooperative Research Centre is a world-leading research organisation that provides thought leadership and break-through technology solutions for capital and insurance markets (www.cmcrc.com). How good is the current method of modelling credit risk for the big banks?
Author(s): Silvio Tarca