How to stress test your credit portfolio?
Article Abstract:
Collecting data over time is the first step in stress testing a portfolio. Once the data is available, preferably two to three years by quarter, a pooled cross sectional time series regression model is estimated. A baseline forecast is then derived by multiplying the model's coefficients by the projected values of the entire explanatory variables over time. Stress testing the credit portfolio is accomplished by replacing the values for the independent variables in the equation and recomputing the forecasts.
Publication Name: Journal of Business Forecasting
Subject: Business, general
ISSN: 0278-6087
Year: 1999
User Contributions:
Comment about this article or add new information about this topic:
Forecasting nonpayment behavior of customers for a mail order company
Article Abstract:
Study results demonstrate that Tobit regression can outperform neural networks and ordinary least squares in the prediction of customer nonpayment behavior. However, Tobit regression achieved its greatest relative accuracy advantage near the lower limit of observations, where most observations were clustered. Tobit regression achieved a 23.47 mean absolute deviation, compared to 25.06 for ordinary least squares, which placed second.
Publication Name: Journal of Business Forecasting
Subject: Business, general
ISSN: 0278-6087
Year: 1996
User Contributions:
Comment about this article or add new information about this topic:
Introduction to Survival Analysis in business
Article Abstract:
The probability of a certain event to occur is provided by the survival model that also alerts a company whether or not a specific account needs a special treatment. The Survival Analysis can be used effectively in retaining existing customers and acquiring new ones in various industries including telecommunication, banking and finance.
Publication Name: Journal of Business Forecasting
Subject: Business, general
ISSN: 0278-6087
Year: 2004
User Contributions:
Comment about this article or add new information about this topic: