In a recessionary environment, companies are looking for ways to accurately predict revenues and losses as unemployment increases and consumers and businesses default on their payment obligations. This paper presents a case study which compares two approaches to predicting losses using econometric methods.

This paper introduces the reader to a technique know as 'Tobit Regression' as a way of modeling a variable with data clustered around a lower limit such as zero. Sorry, this paper is unavailable at this time.

This paper illustrates how you can more effectively determine which customers will respond to new product offers. The same approach can be used for developing new target marketing lists as are used by the credit card companies in extending pre-approved offers of credit in the mail.

This paper presents a simple and easy to follow discussion of a modeling process called "CART" (Classification and Regression Trees) which can be used to supplement the forecasting process and help to better understand the relationships in your data.

This presentation discusses the application of new product forecasting techniques in credit scoring. This is a nontechnical paper written to give a broad overview while highlighting the advantages of using a certain statistical tool in the modeling process.

This paper discusses an approach to forecasting credit losses using an econometric technique called "Pooled Cross-Section Time Series Regression". An illustration is given where a company can aggregate its portfolio information at the geographic level and use it to model and forecast future trends.

This paper discusses the development of a modeling system for credit scoring written in SAS /AF.

Article in Marketing News.

This paper introduces how diffusion models can be used for new product forecasting. This nontechnical paper gives a readable illustration of how a product manager can begin the forecasting process within a systematic framework producing results which can be used to justify the roll-out of a new product.

Continuation of previous article in VISIONS publication.

Continuation of previous article in VISIONS publication.

This paper summarizes a step by step procedure for forecasting new product sales from start to finish. Topics include preliminary data analysis, data preperation, basic statistics, hypothesis testing, correlations, regression, dummy variables, transformations, interaction terms, model validation, and forecast simulation.

This discussion centers around how a consulting company which offered financial services developed a new service which allowed companies to stress test their credit portfolios using econometric methods and data at the sub-state geography level.

This paper presents an approach to new product forecasting using the basics of the product life cycle to describe new product purchases over time. Attention is give as to pent-up demand, the inflection point of the life-cycle, and introduces a concept called the "delay-factor". Numeric examples illustrate simple equations to walk the reader through the process.

This paper is slightly more technical than other presentations on the subject as it gives equations for logistic diffusion as well as gompertz curves. It discusses how to forecast under a variety of data conditions - a). when you have no historical data, b). when you have a short partial history, and how to use similar products as analogies if there exists a complete data history of the product.

The Basel II Capital Accord, ready for implementation around 2006, sets out detailed analytic requirements for risk assessment that will be based on data collected by banks throughout the life cycle of the loan. The purpose of Basel II is to introduce a better risk-sensitive capital framework with incentives for good risk management practices. The more sophisticated banks will be able to take advantage of significantly reduced capital requirements, and therefore be more competitive. This article, the first in a four part series, will give a managerial overview of statistical modeling - an area of analytics that is paramount to the Basel framework.

The purpose of Basel II is to introduce a better risk-sensitive capital framework with incentives for good risk management practices. The more sophisticated banks will be able to take advantage of significantly reduced capital requirements, and therefore be more competitive. Access to these new capital approaches will depend on the bank's ability to develop and implement statistical models that have a proven track record over time. This article, the second in a four part series, will discuss some approaches to the validating statistical models required by the new Capital Accord.

As the Basel Capital Accord nears its 2007 implementation date, bankers are examining its detailed analytic requirements for risk assessment based on data to be collected by banks throughout the life cycle of the loan. The aim of Basel II is to introduce a more risk-sensitive capital framework with incentives for good risk management practices. Basel references words like process or systems about 275 times in the 139-page document. This highlights the importance of a systematic perspective to the analytic side of the house - an approach that integrates programming requirements across different environments with standardized methods and procedures.

The purpose of Basel II is to introduce a better risk-sensitive capital framework with incentives for good risk management practices. The more sophisticated banks will be able to take advantage of significantly reduced capital requirements, and therefore be more competitive. Part of these requirements highlights the need for banks to stress test their credit portfolios. The idea of stress testing should be of interest to banks not only because of Basel, but because developing such approaches can provide the institution with the tools necessary to better manage their capital. Earlier articles in this series have focused on the development and validation of PD and LGD models. In contrast, this article will introduce a different modeling approach that uses aggregated data to predict the impact of economic and portfolio changes on bank default losses.

This paper continues the RMA Journal articles on modeling requirements for Basel II focusing on the problem of missing data. A number of approaches in dealing with the problem are presented here including Multiple Imputation. Illustrations use the SAS software package for missing data.

This paper continues the RMA Journal articles on modeling requirements for Basel II focusing on modeling strategies to improve PD and LGD models. Illustrations use the SAS software package.

This paper continues the RMA Journal articles on modeling requirements for Basel II focusing on statistical techniques to improve validation results when dealing with PD models with relatively few defaults. Specific discussions of Bootstrapping and Jackknife Estimation are presented. Illustrations use the SAS software package.

This paper continues the RMA Journal articles on modeling requirements for Basel II focusing on predicting time to default - a field of study more broadly defined as 'Survival Analysis'. Illustrations use the SAS software package and discuss PROC LIFEREG.

This paper concludes the five part series on practical solutions to challenging problems in developing modeling requirements for Basel II. The purpose of this article is to look at a completely different approach to modeling PD (probability of default) and LGD (loss given default) – neural networks. The motivation for examining approaches other than the traditional techniques recommended so far is because there may be some instances where credit and repayment relationships are more complex.

As the field of credit scoring is focused on predicting 'if' an account will become delinquent over a certain span of time, Survival Analysis can tell us 'when.' Survival Analysis is called different things in different industries event history analysis, reliability analysis, time to failure, and even duration analysis. The purpose of this article is to give an introduction to the subject with an emphasis on how it can be used in banking and finance.

This paper describes the development of a hybrid modeling platform designed especially for that purpose using VB6 and SAS's OLE Automation capabilities. The discussion will revolve around the motivations for such a platform, its features and benefits, some example code and screen captures, and its outlook for the future.

Predicting claim behavior is of considerable interest in the insurance and financial communities. When companies wish to predict whether or not an individual will file at least one claim over a period of time, binary classification models are often chosen. Sometimes, there is a need to go beyond the classification process and predict the actual number of claims. The purpose of this paper was to examine five approaches to modeling insurance claims: 1). OLS, 2). Poisson Regression, 3). Tobit Regression without prior knowledge, 4). Tobit Regression with prior knowledge, and 5). Neural Networks. Sorry, this article is currently not available.

Translation of RMA Article describing Basel II Requirements.

Translation of RMA Article describing Basel II Requirements.

Translation of RMA Article describing Basel II Requirements.

Translation of RMA Article describing Basel II Requirements.

Translation of RMA Article for Basel II Requirements

Translation of RMA Article for Basel II Requirements

Translation of RMA Article for Basel II Requirements

Translation of RMA Article for Basel II Requirements

Translation of RMA Article for Basel II Requirements

The paper shows how you can use the lessons learned in developing credit scoring models in the world of Marketing. For example, common approaches for growing your business such as response models as well as lookalike models are discussed. Also discussed are how to develop cross sell models along with using Survival Analysis in marketing.

The paper shows how you can use the lessons learned in developing credit scoring models in the world of Marketing. This is the same as the above publication, but reprinted in an international Journal - Credit Technology - which has a translation in Portuguese.

This paper discusses some typical problems in deciding which variables to use in regression analysis. Topics discussed are stepwise selection, variable clustering, principle components, and partial least squares. This paper is presented in both English and Portuguese.

This paper discusses how to go about building your own clustering solution using SAS and offers some practical advice along the way. This paper is presented in both English and Portuguese.

In this uncertain environment, institutions should adjust their lending strategies to accommodate for relative risk at the state, MSA, and county levels.

We are in the midst of a severe recession. Some are even venturing to call it the Great Recession. If only we had known! Where were the warning signs? Can the future be predicted? Would financial institutions have benefited from an in-house forecasting function highlighting potential weaknesses?

With today’s economic landscape and the advent of Basel II requirements, risk practitioners are struggling to integrate economic time-series data with their existing credit-scoring models to better understand portfolio risk.

In this article, we examine the individual relationships between liquidation and external factors, both economic and credit, along with their use in a portfolio-level econometric forecasting model. Second, we will present evidence that the use of aggregated credit information at the ZIP code level adds considerable value when prioritizing accounts for maximum collection effectiveness.