Bond volatility threatens banks’ Value at Risk models
June 4, 2013 Leave a comment
June 3, 2013 5:32 pm
Bond volatility threatens banks’ Value at Risk models
By Tracy Alloway in New York
Recent volatility in some of the world’s biggest bond markets could upend the complicated mathematical models that underpin large banks’ trading businesses, risk managers have warned.
“Value at Risk,” or VaR models, have been a key part of banks’ risk management toolboxes for the past two decades, despite being heavily criticised for failing to predict the large losses incurred during the recent financial crisis.The models attempt to forecast how much money banks could lose from trading, within a certain timeframe, by overlaying historic market movements with statistical analysis. Banks typically place internal trading limits based on their VaR estimates, meaning that if they breach the figures they may have to sell off positions.
One large investment bank has already breached its internal risk limits after recent statements from Haruhiko Kuroda, the new governor of Japan’s central bank, people familiar with the situation said. Prices ofJapanese government bonds have been see-sawing in the weeks since Mr Kuroda assumed office.
Some risk managers caution that the models are one tool among many for gauging potential trading risks, meaning a breach of VaR limits might not necessarily translate into large losses for the banks. If banks are regularly exceeding their VaR figures, however, they could risk the ire of regulators.
Traditional VaR models have also been somewhat superseded by “stressed VaR”, along with other key components of Basel banking regulators’ recent reforms.
Banks must now incorporate stressed VaR into their regulatory capital calculations, which means they should in theory be prepared for the worst market volatility from recent years. Stressed VaR is typically twice the amount of “normal” VaR.
However, the recent rise in US and Japanese government bond yields has caused concern among some risk managers. A sudden spike in the yield of Japanese government bonds a decade ago led banks to sell off the debt, in an episode that eventually became known as the “VaR shock” of 2003.
Another shock “could very likely happen for the banks and institutions relying purely on VaR for their signal of risk-taking”, said Denny Yu, head of risk at analytics company Numerix, and a former senior adviser at RiskMetrics, which helped develop VaR.
Mr Yu estimates that before 2008 more than 50 per cent of banks and other financial companies relied on VaR to gauge their risk. “Now that figure is more like 20 per cent, although, if you were being pessimistic you could say it’s more like 40 per cent.”
The recent spurt of market volatility comes just after the end of the first quarter, when many banks reported their lowest potential loss estimates in about six years.
The average VaR estimate of big US investment banks slumped to $82m in the first quarter, according to data compiled by the Financial Times. That means banks would roughly expect to lose more than $82m in a single day very infrequently.
That figure is down from $268m in the first quarter of 2009, in the midst of the financial crisis. VaR averaged about $73m in the five years before 2008.
