Size not the prize in bank stress testing, researchers say

LONDON Thu Aug 2, 2012 10:39am EDT

LONDON Aug 2 (Reuters) - As banking regulators around the world scramble for tools to avert the next financial crisis, Swiss researchers have devised a method that could allow real-time monitoring of stresses in the financial system.

The tool, which a team at Zurich university Eidgenössische Technische Hochschule are calling DebtRank, challenges the idea that the biggest banks are the ones that need closest supervision.

"There is more to systemic importance than asset size," Stefano Battiston, who led the study, told Reuters. "You can easily have two banks with the same balance-sheet size that have a completely different impact.

"We are increasing the granularity in the understanding of who is systemically important," he said.

Using algorithms similar to those that rank results for web pages in a Google search, the researchers argue that even smaller banks can be the source of major stresses in the financial system.

The financial crisis that started in 2008 exposed the limits in national regulation of a banking industry that was globally interconnected.

For much of the crisis, regulators and central banks knew that the financial exposure of banks to each other could act like a lightning rod for transmitting stress around the globe. But the exposures were too opaque to get a detailed picture of what was going on.

Even today, the researchers say, "there is no widely accepted method for working out which institutions in a network are the most important to the stability of the system".

The stress tests that regulators in the U.S. and Europe have periodically run in the wake of the crisis have largely focused on the size of bank balance sheets and their vulnerability to the default of a major debtor.

Battiston and his team argue that this is too binary for the close supervision needed to pick up stresses early enough to prevent the collapse of a bank and the damaging ripple effect through the rest of the system.

"Much attention in the public discussion is currently devoted to the so-called too-big-to-fail institutions," the researchers said.

"Our work shows that this debate should be broadened to the network-theory notion of 'too-central-to-fail'."

Battiston and his team used data from the $1.2 trillion in emergency loans made to global banks by the US Federal Reserve between 2008 and 2010.

Because detailed information on debt exposure was hard to come by, the researchers used interbank equity investment as a proxy, which previous studies have shown to follow a similar pattern.

With the right inputs, the DebtRank tool can generate a number for the total economic loss associated with any given stress situation but can also produce graphics that show how central a bank is in any developing crisis.

One of those graphics is a spiral diagram. As a bank moves to the centre of the spiral it is becoming more systemically important. If it moves to dead centre it suggests there is a risk that could cause a systemwide collapse.

The EU-funded research, which appears in the academic journal Scientific Reports, has excited interest from some of the world's key central banks.

"We are already discussing with the European Central Bank as well as several national central banks about how to apply DebtRanks to their data," they said.

Comments (0)
This discussion is now closed. We welcome comments on our articles for a limited period after their publication.