Why Agency MBS Excess Returns Can Lag OAS Spreads

Summary

The document asks why long-horizon realized excess returns on agency mortgage-backed securities may fall below their average option-adjusted spread over duration-adjusted Treasuries. The response attributes the gap to borrower prepayment behavior changing over time and to prepayment models incorporating those changes only with a delay. Models calibrated to older behavior may therefore misstate the value of the embedded borrower option.

A second mechanism is a negative price difference between mortgage pools valued with newer and older prepayment models under the same OAS, volatility, and interest-rate assumptions. The answer says that an OAS computed using an older model can overstate realized returns. It points to an older Lehman document for the explanation, but provides no citation details, data, decomposition, or test of the effect. The discussion is specific to agency MBS and model evolution, so the stated causes should not be assumed to explain every spread-return gap.

Key ideas

• Borrower prepayment behavior can become more effective over time.
• Prepayment models calibrated on older borrower data may reflect behavioral changes with a lag.
• Pool valuations can differ between newer and older prepayment models under otherwise fixed assumptions.
• An OAS based on an outdated prepayment model may overstate realized excess returns.

Tags

# Realised excess return on agency mbs vs oas spread


# Realised excess return on agency mbs vs oas spread












Would anyone know why realised MBS (esp. Bloomberg Barclays US Agency MBS index) excess returns over duration adjusted Treasury index are systematically below the average OAS spread for long holding periods, let's say for 10y, 20y or 30y? I calculated that annualised realised excess return is roughly 20bp less then average OAS over the same time horizon.

## Answer by Daniel (score 1, accepted)

https://quant.stackexchange.com/a/85687

I found a good explanation in an old Lehman document by Arora, Heike and Mattu. Basically it is because of two reasons: 1) borrowers become more effective in exercising their prepayment option overtime which BAM (Bloomberg Agency Mortgages prepayment model) and other prepayment models that are calibrated on old borrower behavior data take into account only with a lag. 2) Price differential for a pool that is valued with the latest and older version of a prepayment model under the same OAS, vol and interest rate assumptions is usually negative meaning that OAS calculated with old prepayment model overstates realised returns.

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.