Bond Lab Technologies - Modeling Mortgage Default and  Sizing Mortgage Credit Enhancement from the GSEs to the Private Label Market.

The slideshow above outlines the steps taken to determine the appropriate g-fee:

  • First, 1,000 short rate and home price paths are simulated.
  • Next, the home price paths are passed to the Bond Lab default model generating a default curve for each simulated trajectory. Note: the voluntary repayment rate is zero CPR.
  • The default trajectories are applied to the pool of borrowers to determine the expected cumulative loss. Note: cumulative losses are expressed as a percentage of the pool's original balance.
  • The expected cumulative loss of the pool is 3.46% and the pool's expected average life, absent voluntary repayments, is 17.9 years. If one were to apply a 3x loss coverage multiplier then the credit enhancement for this pool is 10.4%. Given the pool's expected average life of 17.9 years the annual g-fee is 58 basis points.
  • The distribution of cumulative losses suggests that in the majority of cases expected cumulative losses are below 5.0%. However, tail losses may exceed 10.0%.


Summary

  • The analysis above implies an ongoing agency g-fee of 58 basis points.
  • In the case of a private label securitization collateralized by prime fixed-rate loans the implied triple-A credit enhancement level is 10.4%. Consequently, the currently used prime mortgage credit enhancement level of between 8.0% and 9.0% are sufficient to protect the triple-A investor against tail risk losses at the 99% confidence level.

Structured Finance Analytics 

The Agency g-fee should be between 55 and 60 basis points before credit adjustments.


In this example we assume a pool of loans whose underlying borrowers all share the same characteristics.


  • 30-year term
  • 30% debt-to-income ratio
  • 720 credit score
  • 25 bps spread to the prime mortgage rate at origination (SATO)


For the analysis we will use the Bond Lab default model and assume a 35% severity.