LoanDynamicsTM is a loan-level model designed to help investors, servicers, insurers, GSEs, and lenders evaluate and quantify the credit risk of their mortgage assets. LoanDynamicsTM forecasts a monthly vector of prepayments, defaults, delinquencies, losses and severities at the collateral level, which can be used for cash flow projections, bond pricing/valuation, loan loss reserves, Asset/Liability management and hedging.
Unlike a full-blown roll rate matrix, LoanDynamicsTM uses a reduced number of transitions—only those transitions with strong economic rationale-- for predicting prepayments, defaults and losses. The LoanDynamicsTM Model uses four delinquency states as shown in the diagram below (Current, Delinquent , Seriously Delinquent and Terminated). These states were chosen because they align with delinquency triggers relevant to bond investors, capture the economics of borrower behavior, and minimize the compounding modeling error introduced when transitions are broken into smaller segments. One of the principal benefits of a reduced form transition matrix is that investors can develop insights more easily about the transitions that contribute to broad performance measures such as 60+ delinquencies, CPR and CDR.
LoanDynamicsTM is unified across credit sector (Jumbo Prime, Subprime, Alt-A/B, High LTV) and product type (fixed, ARM, hybrid, IO, first and second lien) and does not require users to select a credit-specific model. Since the model relies on loan level characteristics as inputs, it produces consistent results across credit sector and product.
The model’s open architecture provides users the flexibility to tune model parameters in order to reflect a view about specific collateral performance or market conditions, which is particularly valuable in a period of economic volatility. In the latest version of LoanDynamicsTM, users can specify a time period for which tuning multipliers remain in effect.
Model Inputs
Term
Coupon
Original Balance
Original LTV
Combined Original LTV
Lien Position
FICO
Delinquency Status
Loan Purpose
Occupancy
Documentation
Property Type
Geographic Location
ARM Characteristics
Economic Drivers
In addition to loan characteristics, the model also requires that users provide an interest rate and HPI forecast. The model forecasts are conditional upon the interest rate and HPI assumptions.
Model Outputs
CPR
CDR
Loss Severity
Cumulative Loss
Cumulative Prepayments
Cumulative Defaults
Monthly Transition Rates
Percentage of Loans in C, D, S & T States
Model Delivery
We offer the LoanDynamicsTM Model as a C+ subroutine, delivered as a Windows dynamically linked library (DLL), a Unix shared object or easy to use application embedded in Excel or Quantrix. The subroutine is ready for integration into customers’ proprietary systems.
Model Description
LoanDynamicsTM is a loan-level model designed to help investors, servicers, insurers, GSEs, and lenders evaluate and quantify the credit risk of their mortgage assets. LoanDynamicsTM forecasts a monthly vector of prepayments, defaults, delinquencies, losses and severities at the collateral level, which can be used for cash flow projections, bond pricing/valuation, loan loss reserves, Asset/Liability management and hedging.
Unlike a full-blown roll rate matrix, LoanDynamicsTM uses a reduced number of transitions—only those transitions with strong economic rationale-- for predicting prepayments, defaults and losses. The LoanDynamicsTM Model uses four delinquency states as shown in the diagram below (Current, Delinquent , Seriously Delinquent and Terminated). These states were chosen because they align with delinquency triggers relevant to bond investors, capture the economics of borrower behavior, and minimize the compounding modeling error introduced when transitions are broken into smaller segments. One of the principal benefits of a reduced form transition matrix is that investors can develop insights more easily about the transitions that contribute to broad performance measures such as 60+ delinquencies, CPR and CDR.
LoanDynamicsTM is unified across credit sector (Jumbo Prime, Subprime, Alt-A/B, High LTV) and product type (fixed, ARM, hybrid, IO, first and second lien) and does not require users to select a credit-specific model. Since the model relies on loan level characteristics as inputs, it produces consistent results across credit sector and product.
The model’s open architecture provides users the flexibility to tune model parameters in order to reflect a view about specific collateral performance or market conditions, which is particularly valuable in a period of economic volatility. In the latest version of LoanDynamicsTM, users can specify a time period for which tuning multipliers remain in effect.
Model Inputs
Economic Drivers
In addition to loan characteristics, the model also requires that users provide an interest rate and HPI forecast. The model forecasts are conditional upon the interest rate and HPI assumptions.
Model Outputs
Model Delivery
We offer the LoanDynamicsTM Model as a C+ subroutine, delivered as a Windows dynamically linked library (DLL), a Unix shared object or easy to use application embedded in Excel or Quantrix. The subroutine is ready for integration into customers’ proprietary systems.