Span 2: a fine balance

Like all financial models, clearing house margin frameworks are heavily scrutinised by regulators, who cast their gimlet eye over any changes the central counterparty wants to make. Unsurprising, given major overhauls have the potential to impact the smooth and safe running of markets. But unlike other financial models, those used by CCPs are also watched equally carefully by their members and clients.

This is doubly the case at CME Group, the world’s largest futures exchange, which is in the throes of moving its many thousands of contracts to a customised value-at-risk methodology. The decision is causing a stir in derivatives markets, and no wonder: the introduction of a new margin model always has the potential to create big winners or losers.

CME has said margins won’t change much with the new framework, but market participants are sceptical. There are multiple competing interests at stake: by their own admission, FCMs want margins kept high to minimise risks and maximise revenue – but on the other hand favour a more accurate model with less operational risk.

Clients generally like to see lower margins to bring down the costs of trading, and would therefore welcome a more responsive model that offered those with balanced portfolios a more accurate reflection of their risk. All the while, other CCPs around the globe, 32 of whom license the current model, are watching carefully.

Span, or the Standard portfolio analysis of risk, to give it its full name, has been around since the 1980s – and when one considers the stakes in any switch, it’s perhaps unsurprising the Merc has persisted with the same core approach for more than three decades. It took Eurex, the first major futures exchange to switch to a VAR-based margin model, several years to develop its new framework, gain the necessary approvals and migrate all its product sets.

To be sure, Span had its quirks – but these often served to keep market participants happy: the model’s clunky approach to setting margins on a product-by-product basis, for instance, was considered overly conservative by market participants – but this often had the effect of keeping margins high, to the satisfaction of dealers, while for clients and licensing CCPs, it meant the model was simple to replicate.

By contrast, the new VAR model introduces transparency and is operationally simpler, but could prove to be more complex for less sophisticated players. Gone are the tens of parameters with the original Span: in their place is a self-calibrating model, which will gauge how a portfolio as a whole performs during various scenarios.

While the new VAR model at the heart of the market risk segment of the framework is more automatic and seemingly less manually driven by decisions from CME staff, there are still plenty of knobs for the CME to tweak that could affect margins – the setting of volatility floors, for instance, and correlation offsets between products sets.

A look under the hood of Span 2

Depending on where these come out, the bourse could indeed make sure margins don’t drop too much from current levels. But many smaller, less sophisticated buy-side firms that trade on CME will also want a predictable model they can easily replicate; more moving parts could get in the way of that.

CME has to tread carefully among these competing interests – but it should not be afraid to make decisions that upset some of its stakeholders, if they are made in the spirit of better risk management. As a major clearer, the bourse has its own safety and soundness to think of, not just its bottom line.

CME, and peers such as LME and Ice, who are also building VAR models, will have to keep this in mind: as one senior regulator privately tells Risk.net, “all new models will have to be justified to us”.

Editing by Tom Osborn