Credit Scoring

Wagner Associates has extensive experience in the development and review of credit scoring models. A scoring model assigns a score to borrowers (either an individual or a company) based on available information about the borrower (such as credit history). The score corresponds to the riskiness of the loan, that is, to the likelihood that the loan will have a specified bad outcome such as default or significant delinquency.

Scoring models are developed by statistical analysis of historical data sets. Wagner Associates has broad knowledge of the statistical methods that are used to develop scoring models. Our knowledge extends from the standard methods, notably logistic regression, to more exotic techniques such as classification trees, neural networks, and expert systems. Wagner Associates has the computing hardware and software tools to analyze large data sets and to develop or review scoring models. We have particular expertise with the S-Plus software system, a comprehensive statistical package for data analysis.

We have worked on scoring models for standard retail credit applications including models for mortgages, for credit cards, and for home equity loans. We have also applied scoring model methods in non-standard circumstances. In fact, the statistical methodology of scoring models applies to any circumstance where one wishes to predict the probability of a particular outcome (good or bad) from any available information source. Currently, we are developing a scoring model for an accounting firm to evaluate the engagement risk of new clients. For example, such risk includes the possibility that a client will draw the firm into detrimental lawsuits.

In addition, Wagner Associates has developed patent-pending methods of automated statistical data mining that can run as client-side or server-side generators of credit scoring models.

Please contact consultus@pa.wagner.com to inquire about our consulting services in credit risk.