Covariance of retail loan product performances
The present invention relates to a method that allows managers of retail portfolios to compute performance time series that have been cleaned of marketing impacts, lifecycles, management actions, and seasonality, leaving only the performance changes due to the environment. These normalized series can be used to compute the necessary covariance matrices for portfolio optimization or computing portfolio-level economic capital. The invention applies to any retail product or segment where vintage-level performance time series are being stored.
1. A computerized method for determining a performance metric for a loan portfolio having multiple types of loans to determine the effect of performance on one loan type on other loan types within the portfolio, the computerized method comprising the steps of:
(a) storing loan performance data for said multiple types of loans on a computer readable medium;
(b) reading said stored loan performance data from said computer readable medium and transforming said stored loan performance by way of a digital computer by decomposing said loan performance data into maturation, exogenous and vintage quality components under program control;
(c) removing the effects of transient maturation and vintage quality components from the performance data for each loan type defining steady state performance data by way of said digital computer under program control; and
(d) computing the covariance between different ones of said multiple types of loans based upon a performance metric using said steady state performance data by way of said digital computer under program control in order to evaluate said loan types in terms of portfolio optimization or economic capital calculation.
2. The computerized method as recited in claim 1 , wherein said step (d) includes:
(d) computing a loss rate using said steady state performance data by way of said digital computer under program control.
3. The computerized method as recited in claim 1 , wherein step (d) includes:
(d) computing profitability using said steady state performance data by way of said digital computer under program control.
4. The computerized method as recited in claim 2 , wherein the capital required is determined based upon said loss rate by way of said digital computer under program control.
5. The computerized method as recited in claim 3 , wherein said portfolio optimization is determined based upon said profitability by way of said digital computer under program control.
6. A digital computer system programmed to determine a performance metric for a loan portfolio having multiple types of loans to determine the effect of performance on one loan type on other loan types within the portfolio, comprising:
a computer readable medium for storing loan performance data for said multiple types of loans;
a digital computer for reading said loan performance data from said computer readable medium and transforming said loan performance data under program control by for decomposing the loan performance data into maturation, exogenous and vintage quality components;
said digital computer programmed to remove the effects of transient maturation and vintage quality components from the performance data for each loan type defining steady state performance data; and
said digital computer further programmed to compute a variance between different ones of said multiple types of loans based upon a performance metric using said steady state performance data in order to evaluate said loan types in terms of portfolio optimization or economic capital calculation.
7. The computer system as recited in claim 6 , wherein said digital computer is programmed to compute the covariance for all of said loan types using said steady state data based upon a performance metric.
8. The digital computer system as recited in claim 6 , wherein said performance metric is the loss rate.
9. The digital computer system as recited in claim 6 , wherein said performance metric is profitability.
10. The digital computer system as recited in claim 8 , wherein said digital computer is programmed to compute the capital required based upon said loss rate.
11. The digital computer system as recited in claim 9 , wherein said digital computer is programmed to compute the portfolio optimization based upon said profitability.
12. The digital computer system as recited in claim 11 , wherein said digital computer is programmed to compute an optimal loan portfolio based upon said portfolio optimization.