SOFR-related Proxies and Approximations for Initial Margin Model Purposes

The information below was distributed to a select group of NFA Member swap dealers (SD) in October 2020.

SDs using the Secured Overnight Financing Rate (SOFR) for discounting purposes and/or trading SOFR-based products must properly treat SOFR exposure when calculating initial margin (IM) using ISDA's Standard Initial Margin Model (SIMM). Since the SOFR curve is not directly modelled in the SIMM framework, SDs will need to map to a proxy (i.e., the closest equivalent curve in the IM calculations) in order to include SOFR exposures in SIMM. ISDA has recommended that the overnight index swap (OIS) curve be used as the proxy for the SOFR curve.

CFTC Regulation 23.154(b)(2)(xi) prohibits an SD from incorporating any proxy or approximation used to capture the risks of the SD's uncleared swaps unless it has first demonstrated to the CFTC or NFA that the proxy or approximation is appropriate.  NFA will not object to an SD including SOFR in the SIMM calculations by OIS proxy subject to the following conditions: 

  • The SD notifies NFA in writing via e-mail to that it is implementing the SOFR curve in its risk management system and that it will include SOFR in the SIMM calculation by the OIS proxy;
  • The SD's Model Risk Management (MRM) team has issued either an approval, a preliminary positive opinion, or a waiver on the SOFR implementation and including SOFR in the exposure margined via SIMM, which may include establishing effective compensating controls or an enhanced monitoring framework to ensure the conservativeness of IM calculated using SIMM, as needed; and
  • If the SD's MRM team authorized the SOFR implementation via waiver, the MRM team must establish a timeframe to conduct its in-depth review of the introduced proxy or approximation in the IM framework within 6 months of its implementation in production.

As part of the MRM team's validation activities to permit the SOFR exposure, via OIS proxy, to be included in SIMM, the MRM team must, at a minimum, consider the following:

  • Scope of the approximation for each curve (i.e., new products and/or discounting and affected trade volumes as measured by trade count, notional amount, and risk sensitivity amounts);
  • Impact analysis on the risk profile of the SD's existing portfolios; and
  • Anticipated impact on SIMM ongoing monitoring testing results (backtesting, benchmarking and risks-not-in-SIMM) for current portfolios due to the introduction of the new products and discounting curve changes.

In order to ensure the conservativeness of SIMM calculations, the MRM team must also assess whether compensating controls, thresholds, and/or an enhanced monitoring framework should be added during the full-scope validation. Any compensating controls previously established should also be reviewed by the MRM team.

NFA's review of an SD's implementation of SOFR in its risk management and in SIMM (via proxy) focuses on the MRM team's validation activities.