Regression Analysis

Regulatory agencies now perform a thorough regression analysis to evaluate specific fair lending issues. Regression analysis yields solid conclusions about patterns or practices that allow regulators to pinpoint fair lending issues. Regulators may perform in-depth underwriting or pricing analysis to look for disparate impact and disparate treatment issues.

The Fair Lending Wiz regression analysis evaluates the consistency of credit decisions. Although the program utilizes appropriate and sophisticated statistical analyses, users will be led through a step-by-step “wizard”. This enables use without statistics programming training to create credit models and to then evaluate the outcomes. As a result, Fair Lending Wiz generates a report detailing the marginal files and giving the user an opportunity to review the results in several different formats.

The regression results are, in fact, broken into four categories:

1. Denied & review for those credits with high approval likelihood that were denied,
2. Approved & review for those credits with low approval likelihood that were approved,
3. Approved - Properly classified where the likelihood from the regression is in agreement with the underwriter's approval decision, and
4. Denied – Properly classified where the likelihood from the regression is in agreement with the underwriter’s denial decision.

The statistical results are also available for review in language and terms that are familiar to the user.

HMDA    CRA    Anti-Predatory Lending    Flood Determinations    MCIF    Terms of Use    Privacy Statement    Section 508    Legal Information    Site Map