Methods and apparatus for utilizing a proportional hazards model to evaluate loan risk

• US 7,392,216 B1
• Filed: 09/27/2000
• Issued: 06/24/2008
• Est. Priority Date: 09/27/2000
• Status: Expired due to Fees

• Current Assignee: Genworth Mortgage Holdings LLC (Genworth Financial Incorporated)
• Original Assignee: Ge Capital Mortgage Corporation (General Electric Company)
• Inventors: Wypasek, Christian John, Palmgren, Matt Andrew
• Primary Examiner(s): WEISBERGER, RICHARD C
• Application Number:
  US09/671,005
• Time in Patent Office: 2,827 Days
• Field of Search: 701/40, 705 35- 45
• US Class Current: 705/37
• CPC Class Codes: G06Q 40/02 Banking, e.g. interest calculation or account maintenance credit or loans G06Q40/03
• CPC Class Codes: G06Q 40/04 Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
• CPC Class Codes: G06Q 40/08 Insurance

Abstract

Systems and processes for more accurate mortgage scoring are described. A proportional hazards model is employed in which not only the occurrence of an event, but also the time to an event such as default of a loan, is considered. In this approach, a hazard rate can be viewed as the chance that an observation will experience an event in the next instant. There are two components to the response, and a binary variable is utilized to indicate whether the event was observed or not, and a time variable. As a result, the number of loans used for modeling is greatly increased, and the time it takes to observe the event, a valuable piece of information in itself, is included in the process. In addition, nonlinear effects are advantageously modeled in a continuous fashion using hat functions to map a series of independent variables. This approach typically yields smaller prediction errors near boundary points.

Claims

1. A method for providing an indication of risk of a loan contemporaneously with origination of the loan, the method comprising the steps of:

receiving mortgage loan data for an applicant for a loan, said mortgage data including data regarding occurrence of an event relevant to the loan and also time to the event;

analyzing the received data utilizing a proportional hazards model to take into consideration not only the occurrence of an event relevant to the loan, but also the time to the event;

computing the indication of risk for the loan using a computer with memory;

transmitting the computed default probability for the loan; and

additionally analyzing the received data utilizing a hat function model to allow nonlinear effects to be modeled in a continuous fashion.

2. The method of claim 1 wherein the indication of risk is a probability of default.

3. The method of claim 1 wherein the proportional hazards model is of the form:

h(t|Z)=h₀(t)*exp(β

^TZ), where h(t) is a hazard rate at time t, Z is a vector of covariates, and β

is a vector of regression coefficients.

4. The method of claim 3 wherein the hazard rate represents a risk of default.

5. The method of claim 4 wherein the hazard rate is represented by a binary variable which indicates whether default was observed or not, and a time observed variable.

6. The method of claim 5 wherein the time observed variable is either a time to default or if default did not occur, a time until observation was censored.

7. The method of claim 5 further comprising the step of:

storing in a database the binary variables and the time observed variables for a plurality of past loans.

8. The method of claim 1 wherein an independent variable, X, is mapped to a series of independent variables X_i which meet the constraints that X_i is a continuous variable over the range [0,1] and each X_i is defined by a fuzzy membership function.

9. The method of claim 1 further comprising the step of:

transmitting a report to a potential loan originator including the indication of risk and highlighting a variable or variables recognized as contributing to the computed indication of risk in a substantial way.

10. The method of claim 9 wherein the indication of risk is a probability of default.

11. A system for predicting a default probability of a loan contemporaneously with origination of the loan, the system comprising:

a server receiving mortgage loan data for an applicant for a loan;

the server including a programmed processor operable to analyze the received data utilizing a software based proportional hazards model;

the server further operable to compute the default probability for the loan; and

a communication mechanism to transmit the computed default probability, wherein the server is further operable to analyze the received data utilizing a hat function model to allow nonlinear effects to be modeled in a continuous fashion.

12. The system of claim 11 wherein the proportional hazards model is of the form:

h(t|Z)=h₀(t)*exp(β

^TZ), where h(t) is a hazard rate at time t, Z is a vector of covariates, and β

is a vector of regression coefficients.

13. The system of claim 11 wherein the hazard rate represents a risk of default.

14. The system of claim 13 wherein the hazard rate is represented by a binary variable which indicates whether default was observed or not, and a time observed variable.

15. The system of claim 14 wherein the time observed variable is either a time to default or if default did not occur, a time until observation was censored.

16. The system of claim 14 further comprising:

a database storing the binary variables and the time observed variables for a plurality of past loans.

17. The system of claim 11 wherein an independent variable, X, is mapped to a series of independent variables X_i which meet the constraints that X_i is a continuous variable over the range [0,1] and each X_i is defined by a fuzzy membership function with said mapping stored in a memory.

18. The system of claim 11 further comprising:

means for automatically generating and transmitting a report to a potential loan originator including the computed probability of default and highlighting a variable or variables recognized as contributing to the computed probability of default in a substantial way.