VaR and the Super Senior ABS CDOs
Interesting times
The loss, the first ever quarterly trading loss for the banking industry, follows weak trading results in the third quarter, and reflects unprecedented turmoil in capital markets, particularly for credit trading.
The poor performance of credit trading in the fourth quarter resulted from challenging market conditions across a range of credit trading markets, but most particularly from concentrated exposure to securities backed by subprime mortgages. Large write-downs on these collateralized debt obligations of asset-backed securities (ABS CDOs) reflect rising defaults on subprime mortgages. The losses on these securities, which initially had very strong credit ratings, also reflect extremely illiquid market conditions. Banks incurred an $11.8 billion loss from credit trading in the fourth quarter, compared to a loss of $2.7 billion in the third quarter.
Gnomes mandated VaR: useless
Concentrations in highly rated but illiquid ABS CDOs, as well as non-normal market conditions, caused several large dealer institutions (both bank and non-bank) to incur very significant trading losses in the fourth quarter.
Historically, these [super senior] ABS CDOs had not exhibited significant price variability given their “super senior” position in the capital structure, so measured risk in VaR models was very low. However, rapidly increasing default and loss estimates for subprime mortgages have caused an abrupt and significant reassessment of potential losses in these super senior ABS CDOs. Because VaR models rely on historical price movements and assume normal market conditions, this particular risk measurement tool may not fully capture the effect of severe market dislocations.
OCC’s Quarterly Report on Bank Trading and Derivatives Activities Fourth Quarter 2007
April 3rd, 2008 at 8:32 am
of course VaR is useless. It is usually based on historical price movements or parametric price assumptions. Past performance is no indication of future performance. A poor risk system is worse than no risk system for it leads to a false sense of security and therefore reckless behaviour. I would rather have everone a little cautious driving 60MPH rather than everyone going 95MPH knowing they have airbags. Read the book Normal Accidents if you really want to understand risk mitigation gone wild.
April 3rd, 2008 at 8:39 am
Reminds me of a strike by firemen in England a few years ago, fires number went down by half for a few weeks.
April 4th, 2008 at 3:58 pm
But even better is driving @ 60 mph with airbags. VaR is not of course useless, of course. It one metric on the dashboard, useless in isolation maybe. David Harper
April 4th, 2008 at 4:30 pm
The original idea of VaR, came from a trading desk at Bankers Trust and applied to short term trading in liquid markets it is fine/workable in the right hands. It is the extension of VaR to illiquid assets or derivatives where the distribution of returns is not well known ( I am being nice here) that poses a serious problem. The SEC text says it all, works fine in “normal” times, then crashes at the first bump in volatility because “this particular risk measurement tool may not fully capture the effect of severe market dislocations.” Well what’s the point then ?
April 4th, 2008 at 5:07 pm
“The SEC text” should be the OCC text.