Somewhere in the offices of those international banks which every day seem to write off a few more billion in losses there are teams of mathematical whiz kids using computers to calculate the risks involved in the banking business. Risk modelling and quantitative approaches to risk management and the allocation of capital is what they call it and it is part of the requirements of the Basel II international agreement that is designed to keep world banking sound.

It all reads as a very sensible system until you read stories like the one this morning which tells how a low level equities derivative trader cost the French banking giant Societe Generale $US7 billion on trades that went wrong. That makes the problems of the National Australia Bank a few years ago with its rogue traders pale into insignificance. But don’t think that an out of control punter with the bank’s money is the only risk. Have a look at a report like that by International Monetary Fund economists Manmohan Singh and Mustafa Saiyid who have drawn attention to plenty of other ways that banks can get into trouble.

Messrs Singh and Saiyid explain how global markets turmoil is in part due to a lack of appropriate measures to evaluate the risk of new financial products. Following the collapse of the US subprime market in mid 2007, market worries about the exposure of the structured securities to the subprime crisis caused a credit freeze that made many market players use valuation models that no longer worked in the meltdown, the report said.

In short the figures the mathematical wizards were putting in to their sophisticated computer models had no relation to what the assets being measured would actually bring if sold. “Most banks in the United States,” say Singh and Saiyid, “have not yet marked their assets to genuine transaction prices.” Some market players “increasingly relied on ratings as a measure of default risk and inappropriately compared them to those on plain vanilla corporate debt, which has different sensitivities to market conditions”.They suggest that market participants seek to regularly put a portion of their complex structured securities on the market to obtain a valid valuation.

In Australia, the Australian Prudential Regulation Authority (APRA) which has the task of keeping our financial system sound, is clearly aware of the limitations of the modelling techniques it oversees. Its annual report for 2007 noted that the use of ‘net income surplus’ models in place of more traditional lending criteria has led to higher lending for housing against given levels of borrower income. “However, the key assumption of these models — that all net income above basic living expenses and other fixed commitments is potentially available to service the housing loan — has not been tested in adverse economic conditions,” said the report. It continued:

For this reason, boards and management of [banks] need to ensure that they understand the construction of the debt-serviceability models they are using, the quality of the data inputs, the sensitivity to parameters selected and the performance of the models over time. APRA’s review has provided peer group data that enable it to direct its supervisory attention to those [banks] that appear to have the more aggressive lending practices.

Not that it appears Australians have too much to be concerned about as the impact of the US banking crisis spreads around the world. As part of a recent IMF assessment of the Australian financial system, the Australian authorities and the five largest banks, subjected the banks to a stress test intended to measure their resilience to a large fall in various asset prices, a large increase in their wholesale funding costs because of a sharp deterioration in investor sentiment, and a recession. The findings as outlined at a recent Reserve Bank conference were:

Among the various changes assumed to occur over the course of 2006 were: a 30 per cent fall in house prices; a significant depreciation of the exchange rate; higher wholesale funding costs for banks and unchanged official interest rates; a short recession in which real GDP falls by 1 per cent, driven by an unprecedented 2½ per cent contraction in household consumption; and an increase in the unemployment rate and inflation rate of about 4 percentage points and 2¼ percentage points, respectively. In aggregate under this scenario, the banks reported a fall in profits relative to the second half of 2005 of around 40 per cent after 18 months, easing to 25 per cent after another 18 months – although there was considerable variation in the individual banks’ results, none of the fi ve banks reported a loss during the forecast period. Around one-half of the fall in aggregate profits came from an increase in bad debt expenses: households’ cutbacks in consumption in order to continue servicing housing loans (together with prepayment buffers and mortgage insurance) restrained the losses on housing loans but contributed to the general slowdown in the economy, which resulted in an increase in business bad debt expenses. Most of the remaining fall in profi ts was due to increased funding costs and lower income from wealth management operations. While there were some caveats associated with the exercise, one of which was the surprisingly large variation in outcomes across banks, on balance we consider the results of this stress test as providing support for our assessment that the banking system in Australia is well placed to withstand a major adverse shock.