.jpg)
Has SPAN outlived its usefulness, or does the constant stream of new releases and updates make it still useful after 25 years? (And why did CME abbreviate Standardized Portfolio Analysis of Risk as SPAN instead of SPAR?)
SPAN runs 16 scenarios in its standard calculations, because it was limited by the computing power available in 1988. It has become an industry standard, used by CME and 50 registered exchanges and clearing organizations around the world. But not by Eurex which has developed a Value-at-Risk (VaR) model.
Eurex has developed its own margin calculation engine, said Matthias Graulich, EVP and global head of OTC development and business relations at Eurex. Its aim is to make sure the clearing house has sufficient funds to take over and liquidate positions if a member firm fails. SPAN’s 16 scenarios are not enough to cover all the different dimensions of risk in the markets, especially when swaps are included because they are more complex and require looking at more risk factors.
Eurex also takes advantage of modern technology and today’s computational speed and capacity.
“We decided an integrated approach is important. We saw margin requirements increasing. You need to look at the risks of these positions as a portfolio. We consider 1,000 scenarios and 25 risk factors. It is only possible with the computational power we have these days. We calculate within seconds the incremental risk of a new trade.”
Eurex has different liquidity horizons for different instruments, like four days for equities he added.
“We believe that is achievable with the liquidity in the market. You need different liquidation horizons for different products.”
The VaR approach rewards clients with very balanced portfolios, he added. They can see significant reductions in margin requirements.
SPAN can handle multiple liquidation periods, said Matthew Waldis, executive director for risk management at the CME Group. It has provisions for liquidity and more work in that area is in development. Other improvements coming include more complex algorithms for more efficient netting and expansion beyond 16 scenarios, he added.
SPAN has done yeoman’s work, said Carl Gilmore, CEO at KCG Futures, using terminology that suggests medieval times. “But as everyone who uses it knows, there are things it hasn’t done particularly well; one is to accurately asses the risk for tail events. Risk-based margining models make it tremendously difficult to estimate the cost of liquidation. You want enough collateral to liquidate a position of any size and you will have to do it under the worst circumstances. CME has done a great job of updating SPAN over the years, but it will need to go in a different direction.”
Markets are inter-connected today in a way that they weren’t in 1988. His firm uses another risk management tool to stress its positions and compare the results to SPAN, he said, but he would like to see exchanges agree on standards or a single approach to reduce manual work reconciling the results of multiple models.
“We have to manage all that, and we are in a zero tolerance regulatory environment. If we make a mistake, we are the ones wearing the regulatory risk, much more so than the clearing houses.”
