Using Score to Improve Understanding of Economic Value versus Market Value of IO Strips

Mortgage assets are priced to create an option adjusted spread (OAS) to treasury securities. Different mortgage assets may trade at different OAS levels.  The spread to treasuries is primarily attributed to the funding and other risks created by the borrowers' option to prepay.  Scoring allows for improved understanding of the price and value because it improves the user's understanding of prepayment behavior. To demonstrate this improved value, FIS Applied Analytics recently analyzed an $18.7 billion conforming 30-year mortgage-servicing portfolio. The portfolio contained 169,000 loans. These loans were placed in 105 pools that had similar coupons and origination dates. We then calculated a price for each pool that would generate a 90 basis point OAS on interest only strips of 25 basis points.

Register to receive these case studies

Using Score to Improve the Economics of Servicing Portfolio Retention

We demonstrate the $203,000 cost savings (versus the use of a standard prepayment model) associated with using loan level scoring in the solicitation and retention of existing borrowers. FIS Applied Analytics scored a portfolio of 169,652 loans as of the end of January 2003.  Using the FIS Applied Analytics prepayment model, as modified by the score, we forecasted each loan's single monthly mortality for the next three months.  Forecasts of short-term prepayments as a function of a score are also available on the FIS Applied Analytics web site for users that do not license the FIS Applied Analytics prepayment model. We arranged the loans in buckets and ranked them according to the forecast SMM. Each bucket contained 10% of the portfolio or 16,965 loans. Note that the score enables us to understand forecast SMM. This understanding is critical. A mere ranking of prepayment propensity could lead to the solicitation of loans that would not be likely to refinance.

Register to receive these case studies

Loss Avoidance in Pricing Mortgage Servicing Rights via the FIS Applied Analytics Score Product

Here we demonstrate the use of the FIS Applied Analytics prepayment score in the avoidance of a $900,000 loss in a $10 million investment in mortgage servicing rights.

The value of the loan level modifier to a standard model can be demonstrated by putting yourself in the shoes of a servicing buyer:

Assume that:

• You are given the task of pricing the acquisition of $1 billion of mortgage servicing rights
• You do not have a loan level model and the seller does.
• The seller has used his model to select loans that will prepay 30% faster than your model is forecasting.
• You will be responsible for any accelerated amortization of servicing values caused by prepayment speeds that were forecast incorrectly.
• You rely on a prepayment model to calculate the price of servicing based on the option-adjusted spread (OAS) of the embedded interest only strips (IO's).
• You have a target OAS on the IO's of 50 basis points. This is the indicated market spread for the IO strips.

Register to receive these case studies

Loss Avoidance in Hedging Mortgage Servicing Rights (Hedging MSRs)

Here we demonstrate the use of the score to avoid a $10.5 million loss in the hedging of mortgage servicing rights with a value of $150 million.

The value of the loan level modifier to a standard model can be demonstrated by putting yourself in the shoes of an executive that is responsible for hedging the value of mortgage servicing rights.

Assume that:

• You are long a $10 billion servicing rights portfolio valued at $150 million
• Your job is to hedge the portfolio value. Your portfolio value, when combined with your hedge value should remain constant when exposed to a one-sigma change in interest rates, which is plus or minus 60 basis points today.
• In order to hedge your servicing rights value you will acquire assets that have an effective duration that offsets the negative duration of your servicing rights. If rates move up or down, the value of your hedge should offset the value of your MSR's. (We will not address negative convexity in this simplified example).
• You don't have a loan level model. As a result, your projected prepayment speeds are 30% too slow but you won't know this until its too late.

Register to receive these case studies

Related Links