Vectors™ MBS Prepayment Model

Fixed Rate MBS Model
Adjustable Rate MBS Model

Fixed Rate MBS Model

Model Overview: The Vectors^TM MBS Prepayment Model for fixed rate loan types uses agency data though July 2006 and non-agency (jumbo prime and sub prime) data though December 2005 and incorporates two factors, vintage-based loan size effect and yield curve effect, to improve the fit between actual and forecasted prepayments. The agency model can adjust the refi and turnover tuning parameters into unique risk multipliers to capture the impact the extended disclosure data has on prepayments.

Model Inputs: Loan Type, Gross WAC, Original WAM, Loan Age

Loan Level Inputs (Jumbo Prime and Sub Prime only): Loan Size, Original LTV

Enhanced Pool Characteristics for Agency Loan types: WA Original Loan Size, WA Original Credit Score, WA Original LTV, States of Origination, Loan Purpose, Occupancy, Property Type

Factors Considered: Turnover (tuned by enhanced data), Refi (tuned by enhanced data), Burnout, Seasoning, Seasonality, SATO, Home Price Appreciation (state level), Vintage-based Loan Size Effect and Yield Curve Effect.

Functional Form: Active-Passive Decomposition with naturally occurring Burnout

Loan Types Covered:

Delivery: Subroutine and Online Demo (request a demo)

Performance Reports: Available. (View Performance Reports)

Fixed Rate Tuning Parameters (for Agency loan types only) : Scaling, Turnover, Refi, Slide, Burnout, Age, Lag, Credit Cure, SATO, Cashout, Curve at Origination (CATO)

Enhanced Data & Risk Multipliers: The Vectors™ MBS Prepayment with risk multipliers uses adjustments to the Refi and Turnover tuning parameters to account for the impact of extended data items.

Model Documention:
Fixed-Rate Agency MBS Prepayments & Model Enhancements
Fixed-Rate Home Equity Loan Prepayment Model
Divide & Conquer: Exploring New OAS Horizons, Parts I, II & III

Adjustable Rate MBS Model

Model Overview: The Vectors^TM MBS Prepayment Model for adjustable rate loans brings a brand new suite of ARM models into ADCo’s product line. Based on agency ARM data though July 2006 and non-agency (jumbo prime and subprime) ARM data though December 2005, the model incorporates the Active-Passive Decomposition method of modeling burnout found in the fixed rate model, and it utilizes a new “risk aversion” methodology to model prepayment behavior near ARM reset dates. The risk aversion refers to the borrower’s propensity to refinance before reset because of their perception of a substantially higher interest rate at the reset date. The model uses a forward averaging methodology which forecasts the expected WAC the borrower will face at reset, based on the cap/floor/margin structure of the loan.

This model also has the ability to handle reset behavior past the initial roll. An important feature is that the user does not select a specific model, like “FNMA_3YR_HYBRID.” Model selection is done based on the user inputs into the model, which are listed below.

Other major featuers of the model include the ability to forecast prepayments for Interest Only (IO) loans and Prepayment Penalty loans, as well as allowing the user to choose the interest rate on which the ARM is indexed. The ARM model does not yet consider the Enhanced Data. Enhanced Data, however, is incorporated in the fixed-rate models.

Model Inputs: Original Gross WAC, Current Gross WAC, Original WAM, Loan Age, Initial Reset Months, Months Between Reset, IO Months, Penalty Months, ARM Index, Gross Margin, Life Cap, Life Floor, Periodic Cap, Periodic Floor.

Loan Level Inputs (Jumbo Prime and Sub Prime only): Loan Size, Original LTV

Factors Considered: Turnover, Refi, Burnout, Seasonality, Risk Aversion, SATO, Home Price Appreciation (Cashout) and Yield Curve Spread Effect

Functional Form: Active-Passive Decomposition with Risk Aversion

Loan Types Covered:

Delivery: Subroutine and Online Demo (request a demo)

Performance Reports: Available for Conventionals (view Performance Reports)

Tuning Parameters: Turnover, Refi, Cashout, Credit Cure, Aging, SATO, Curve Spread, Lag, Scale, Slide, and Burnout

Risk Multipliers & Enhanced Data: The Vectors™ MBS Prepayment model for agency loan types supports use of enhanced data fields (Loan Size, FICO, LTV, etc.) for fixed rate models only. The next release of the model will include these factors for all ARM models. The Non Agency Prepayment Model for Jumbo Prime and Sub Prime loan types uses risk multipliers for Original Balance and Original LTV.