Method and system for assessing loan credit risk and performance

• US 8,433,631 B1
• Filed: 01/28/2011
• Issued: 04/30/2013
• Est. Priority Date: 09/11/2003
• Status: Active Grant

• Current Assignee: Fannie Mae
• Original Assignee: Fannie Mae
• Inventors: An, Mark Y., De Ritis, Cristian A., Rosenblatt, Eric L.
• Primary Examiner(s): Colbert, Ella
• Assistant Examiner(s): Trotter, Scott S
• Application Number:
  US13/016,826
• Time in Patent Office: 823 Days
• Field of Search: 717/104, 705/35, 705/36.R, 705/38
• US Class Current: 705/35
• CPC Class Codes: G06Q 40/00 Finance; Insurance; Tax strategies; Processing of corporate or income taxes
• CPC Class Codes: G06Q 40/02 Banking, e.g. interest calculation or account maintenance credit or loans G06Q40/03
• CPC Class Codes: G06Q 40/03 Credit; Loans; Processing thereof
• CPC Class Codes: G06Q 40/06 Asset management; Financial planning or analysis

Abstract

A method and system for evaluating the performance of a study group of loans relative to the performance of a control group of loans is presented. The described method includes determining a study group and a control group from a portfolio of individual loans, determining a set of control variables associated with each of the individual loans included in the study group and the control group, partitioning the control group into one or more cells based on the set of control variables, calculating an estimated performance value for each of the individual loans in the study group based on the cells; and calculating a performance difference for the study group. The performance difference has a standard normal distribution with zero mean which allows statistical significance to be estimated using standard probability tables. The control group may either include or exclude the individual loans in the study group to eliminate inaccuracies.

Claims

1. Machine-readable non-transitory media having stored therein a set of instructions that when executed cause a computer to implement a process for evaluating the performance of a study group of loans relative to the performance of a control group of loans, the process comprising:

determining a study group from a portfolio of individual loans, each of the individual loans in the portfolio having a performance variable associated therewith and a corresponding loan-specific value thereof, wherein the loan-specific performance variable value indicates delinquency and/or default of an individual loan;

determining a control group from the portfolio of individual loans;

determining a set of control variables associated with each of the individual loans included in the study group and the control group, wherein at least one control variable in the set of control variables indicates an estimated loan-specific performance value;

partitioning the control group into one or more cells based on the set of control variables, wherein each cell includes one or more loans with similar loan specific values for each of the control variables in the set of control variables, and wherein each individual loan in the study group shares similar values with at least some of the control variables associated with each loan in at least one corresponding cell from the one or more cells;

calculating an estimated performance value for each of the individual loans in the study group based on the control variables values indicating estimated loan-specific performance associated with loans of the at least one corresponding cell associated with the individual loan in the study group;

calculating a performance difference for the study group, wherein the performance difference comprises the difference between an average of the loan-specific performance variable value for each individual loan in the study group and an average of the estimated performance value for each individual loan in the study group; and

storing the calculated performance difference in a database.

2. The machine-readable non-transitory media of claim 1, wherein the control group is inclusive of the study group.

3. The machine-readable non-transitory media of claim 1, wherein the control group and the study group are mutually exclusive.

4. The machine-readable non-transitory media of claim 1, wherein the performance variable is a binary variable.

5. The machine-readable non-transitory media of claim 1, wherein the performance variable is a continuous variable.

6. The machine-readable non-transitory media of claim 1, wherein the estimated performance value for each individual loan in the study group comprises the average of the loan-specific performance variable values associated with each of the individual loans in the corresponding cell.

7. The machine-readable non-transitory media of claim 1, wherein an imported prediction model is used in partitioning the control group and in calculating the estimated performance value for each of the cells.

8. The machine-readable non-transitory media of claim 1, wherein the process further comprising the step of testing the statistical significance of the performance difference.

9. Machine-readable non-transitory media having stored therein a set of instructions that when executed cause a computer to implement a process for evaluating the performance of a study group of loans relative to the performance of a control group of loans, the process comprising:

determining a study group, the study group comprising a first grouping of loans having a common grouping characteristic and being selected from a portfolio of individual loans, each of the individual loans in the portfolio having a performance variable associated therewith and a corresponding loan-specific value thereof, and one or more control variables associated therewith and corresponding loan-specific values thereof, wherein the loan-specific performance variable value indicates delinquency and/or default of an individual loan;

determining a control group, the control group comprising a second grouping of loans selected from the portfolio of individual loans;

determining a set of control variables, the set of control variables comprising one or more of the one or more control variables associated with each of the individual loans included in the study group, wherein at least one control variable in the set of control variables indicates an estimated loan-specific performance value;

partitioning the control group into one or more cells, each cell comprising the individual loans in the control group having the same or similar loan-specific control variable values for each of the control variables included in the set of control variables, wherein each individual loan in the study group shares similar values with at least some of the control variables associated with each loan in at least one corresponding cell from the one or more cells;

calculating an estimated performance value for each of the individual loans in the study group based on the control variables values indicating estimated loan-specific performance associated with loans of the at least one corresponding cell associated with the individual loan in the study group;

calculating a performance difference for the study group, the performance difference comprising the difference between an average of the loan-specific performance variable value for each individual loan in the study group and an average of the estimated performance value for each individual loan in the study group; and

storing the calculated performance difference in a database.

10. The machine-readable non-transitory media of claim 9, wherein the control group is inclusive of the study group.

11. The machine-readable non-transitory media of claim 9, wherein the control group and the study group are mutually exclusive.

12. The machine-readable non-transitory media of claim 9, wherein the performance variable is a binary variable.

13. The machine-readable non-transitory media of claim 9, wherein the performance variable is a continuous variable.

14. The machine-readable non-transitory media of claim 9, wherein the estimated performance value for each individual loan in the study group comprises the average of the loan-specific performance variable values associated with each of the individual loans in the corresponding cell.

15. The machine-readable non-transitory media of claim 9, wherein an imported prediction model is used in partitioning the control group and in calculating the estimated performance value for each of the cells.

16. The machine-readable non-transitory media of claim 9, wherein the process further comprising the step of testing the statistical significance of the performance difference.

17. A computer-implemented system configured to evaluate the performance of a study group of loans relative to the performance of a control group of loans, the system comprising:

memory including loan variable data for a portfolio of individual loans, wherein the load variable data includes a performance variable for each of the individual loans and a corresponding loan-specific value thereof, and wherein the loan-specific performance variable value indicates delinquency and/or default of an individual loan; and

a processor coupled to the memory and operable to execute programmed instructions, wherein the programmed instructions;

A. determine a study group from the portfolio of individual loans;

B. determine a control group from the portfolio of individual loans;

C. determine a set of control variables associated with each of the individual loans included in the study group and the control group, wherein at least one control variable in the set of control variables indicates an estimated loan-specific performance value;

D. partition the portfolio of individual loans into one or more cells based on the set of control variables, wherein each cell includes one or more loans with similar loan specific values for each of the control variables in the set of control variables, and wherein each individual loan in the study group shares similar values with at least some of the control variables associated with each loan in at least one corresponding cell from the one or more cells;

E. calculate an estimated performance value for each of the individual loans in the study group based on the control variables values indicating estimated loan-specific performance associated with loans of the at least one corresponding cell associated with the individual loan in the study group; and

F. calculate a performance difference for the study group;

wherein the performance difference comprises the difference between an average of the loan-specific performance variable value for each individual loan in the study group and an average of the estimated performance value for each individual loan in the study group.

18. The system of claim 17, wherein the loan variable data includes a loan-specific performance variable value associated with each of the individual loans in the portfolio, and further includes one or more control variables associated with each of the individual loans in the portfolio, and wherein the vector of control variables comprises one or more of the control variables associated with each of the individual loans in the portfolio.

19. The system of claim 18, wherein the loan-specific performance value is a binary variable.

20. The system of claim 18, wherein the loan-specific performance variable is a continuous variable.

21. The system of claim 18, wherein the estimated performance value for each of the individual loans in the study group comprises the average of the loan-specific performance variable values associated with each of the individual loans in the corresponding cell.

22. The system of claim 18, wherein an imported prediction model is utilized in partitioning the portfolio of individual loans into one or more relevant cells and in calculating the estimated performance value for each of the cells.

23. The system of claim 18, wherein the control group is inclusive of the study group.

24. The system of claim 18, wherein the control group and the study group are mutually exclusive.

25. The system of claim 18, wherein the programmed instructions further test the statistical significance of the performance difference.

26. Machine-readable non-transitory media having stored therein a set of instructions that when executed cause a computer to implement a process for evaluating the performance of a study group of loans relative to the performance of a control group of loans, the process comprising:

determining a study group from a portfolio of individual loans;

each of the individual loans in the portfolio having a performance variable associated therewith and a corresponding loan-specific value thereof;

determining a control group from the portfolio of individual loans;

the control group being exclusive of the study group;

determining a set of control variables associated with each of the individual loans included in the study group and the control group, wherein at least one control variable in the set of control variables indicates an estimated loan-specific performance value;

partitioning the control group into one or more cells based on the set of control variables, wherein each cell includes one or more loans with similar loan specific values for each of the control variables in the set of control variables, and wherein each individual loan in the study group shares similar values with at least some of the control variables associated with each loan in at least one corresponding cell from the one or more cells;

calculating an estimated performance value for each of the individual loans in the study group based on the control variables values indicating estimated loan-specific performance associated with loans of the at least one corresponding cell associated with the individual loan in the study group;

calculating a performance difference for the study group;

wherein the performance difference comprises the difference between;

(i) the study group size multiplied by the summation of;

(a) the average values of the loan-specific performance variable multiplied by (b) the number of loan-specific performance variables included in the average value of the loan-specific performance variable and (ii) the study group size multiplied by the summation of;

(a) the average values of the estimated performance value for each individual loan in the study group multiplied by (b) the number estimated performance values included in the average value of the estimated performance value; and

storing the calculated performance difference in a database.