Author Details

Yihai Yu

Yihai Yu

Executive Director, MSCI Research

Zhi-Guo Huang

Zhi-Guo Huang

Executive Director, MSCI Research

Social Sharing

Extended Viewer

MBS Valuation and Risk Management Under SOFR

  • As LIBOR is set to cease, derivatives linked to the secured overnight financing rate (SOFR) have become the market standard and are key components of an option-adjusted valuation framework for mortgage-backed securities (MBS).
  • The valuation benchmark for MBS has evolved around different interest-rate curves and volatility surfaces in the past 30 years. The lack of a Treasury-based volatility surface has been a long-standing hurdle for consistency.
  • We explore the key characteristics of a new unified SOFR-based framework in what we expect to be a seamless LIBOR-SOFR transition for MBS analytics.

With the anticipated LIBOR cessation after June 2023, the transition from LIBOR to new benchmark the secured overnight financing rate (SOFR) is entering a new stage. We analyze this transition’s potential impact in the context of historical valuation frameworks and hedging practices for mortgage-backed securities (MBS). While we expect a smooth LIBOR-SOFR transition for investors, we explore the key characteristics of a new SOFR-based framework for valuing MBS.


Which curve is more correlated to MBS performance?

Duration hedging is a key challenge in MBS investment and risk management. Hedge ratios for MBS traded in the to-be-announced market are based on market consensus — i.e., how much of an interest-rate product will be used relative to MBS holdings to reach neutral duration. Very often, MBS risk management and performance attribution rely on model-based duration measures such option-adjusted duration (OAD), especially for less liquid collateralized mortgage obligations such as inverse interest-only instruments. Hedging such instruments can be a very complicated matter. One basic question is which curve to use for hedging and measuring duration.


How MBS-valuation benchmarking has evolved

One-year rolling regression with the x-axis label centered on the middle of the one-year period. Ten-year Treasury and swap rates are used vs. the 30-year Fannie Mae current-coupon rate. Y-axis was set to start from 0.5, omitting the correlation breakdown for a short period of time near the onset of the COVID-19 pandemic in 2020, to have a more meaningful scale for the other periods. Source: MSCI Agency MBS Current Coupon Marking Model


A brief history of MBS hedging

After the emergence of MBS, hedging with the Treasury curve was standard practice — until early 2000, when the LIBOR-based swaps showed higher correlation than Treasury-MBS performance. This could mainly be attributed to two factors during that period: MBS ownership was skewed toward LIBOR-based investors, such as the government-sponsored enterprises (GSEs) Fannie Mae and Freddie Mac; and Treasury issuance was relatively scarce, which led to technical issues in practice.

But shortly after the 2008 global financial crisis, Fannie Mae and Freddie Mac reduced their MBS holdings drastically, and then the LIBOR scandal contaminated the reliability of LIBOR-based swaps. Treasury-based hedging started to dominate again, as an abundance of Treasury issuance and the Federal Reserve’s quantitative easing/tightening affected both Treasurys and MBS simultaneously and proportionally, leading to higher Treasury-MBS correlation.

Subsequently, investors started to rely more on Treasury-based option-adjusted modeling. But the lack of a Treasury-based volatility surface forces modelers to rely on LIBOR-based swaptions for volatility calibration. Benchmarking to Treasury curves with swaption-volatility calibration certainly creates inconsistency.


Liquidity of SOFR-based swaps and swaptions has improved

According to the recent progress report on the SOFR transition by the Alternative Reference Rates Committee, SOFR-linked linear swaps and nonlinear derivatives have been predominant in the markets.1 The daily average interdealer outright linear swap risk traded is now dominated by SOFR (representing about 90% of market activity since early 2022). Swaptions today are also almost mostly issued under SOFR, instead of LIBOR. This move away from LIBOR has paved the way for the transition to a unified, SOFR-based valuation benchmark for MBS.


SOFR and LIBOR swap curves have converged to a stable spread term structure


Significant difference between the inverted curves of Treasurys and SOFR


Pricing impact for different settings of interest-rate curves

Interest-rate curves can play a part in MBS valuation in multiple ways:

  • Index-linked coupon calculation: This is very straightforward, as the calculation is simply specified by a formula in the securitization’s prospectus, such as one month LIBOR + spread, or SOFR 30-day average + spread.
  • Discount curve: This is the benchmark for the option-adjusted spread that investors calculate.
  • Mortgage-rate projection: The MSCI Mortgage Rate Model projects future mortgage rates based on the rate curve, current-coupon spread and primary-secondary spread. The steeper the curve used, the higher the future mortgage rates generated.
  • Swaption-volatility surface: Used for calibration, it generates the future rate paths to be used for the above three cases. The lack of a Treasury volatility surface typically forces users to apply spread adjustment on top of the LIBOR-based-swaption calibration. But the different curve level, shape and dynamics inherently create inconsistency for this approach.

SOFR and LIBOR curves mostly differ by a relatively simple spread term structure, especially for the longer tenors, which tend to be more important for mortgage-rate derivation. We recalibrated the MSCI Agency MBS Mortgage Rate Model to adjust for this difference. As shown in the table below, the impact on the forward prepayment projection is very minimal.

The volatility surface between SOFR and LIBOR is also minimal, with LIBOR volatility 2 basis points (bps) higher in the measure of normal bp volatility. As SOFR rates are lower by about 20 bps on average, the SOFR option-adjusted spread (OAS) is higher by about 20 bps due to the discount-curve effect. The impact on the key risk measures is minimal, such as OAD, option-adjusted convexity (OAC) and vega (a common measure of volatility risk). This is good news for investors hoping for a smoother SOFR transition.


Pricing impact on a typical MBS, under different curve settings

Estimated impact on the OAS, OAD, OAC and conditional prepayment rates (CPR) of a uniform MBS from the to-be-announced market, with a 30-year 6% coupon, under different curve settings. Source: MSCI Agency MBS Model Suite, MSCI Two-factor Interest Rate Model

When the Treasury curve, calibrated to SOFR volatility, is used for just discounting, but not mortgage-rate projection, prepayment projection will be the same as the mortgage-rate projection is still based on forward SOFR rates in this case. OAS is approximately 20 bps tighter, because the Treasury curve is higher than the SOFR curve at the point near the bond’s weighted average life.

Last, when the mortgage-rate projection is also based on the Treasury curve, which is steeper than the SOFR curve, the prepayment speed is slower due to the higher forward mortgage rate. As a result, OAS is 4 bps wider.

In summary, we have reason to be optimistic about a smooth transition in switching from LIBOR- to SOFR-based risk analytics. Meanwhile, investors who currently rely on the Treasury curve for a benchmark may wish to investigate the differences in SOFR- and Treasury-based approaches to bond-level analytics.



1“ARRC Readout for January 19, 2023 Meeting.” Alternative Reference Rates Committee, Jan. 25, 2003.



Further Reading

Agency MBS in 2023: Uncharted Territory

Managing Against MBS Indexes: A Duration Perspective

MSCI Agency Fixed Rate MBS Prepayment Model Version 2.0

A ‘Normal’ Choice of Interest-Rate Model for MBS