Served as testifying expert in National Fair Housing Alliance, Inc., et al v. The Prudential Insurance Company of America, which concerned the disparate impact of insurance credit scoring. The case involved African Americans allegedly being charged unnecessary risk premiums for homeowners insurance as a result of credit scoring. Analyzed statistical impact of credit-score-based pricing systems on the premiums charged to African Americans. Constructed alternative predictive credit-score models and pricing systems to reduce disparate impact but maintain equivalent level of predictability and profitability.

Provided statistical and economic analysis on behalf of a major financial institution with multiple mortgage business units. Developed statistical models to detect pricing disparities between protected and unprotected classes and to analyze the relationship between its mortgage pricing strategies and credit risk. Currently developing alternate risk-adjusted pricing algorithms.

For a $2 billion retail and broker mortgage originator, analyzed HMDA data to determine source of racial and gender pricing disparities. Investigated branch and loan officer level pricing to uncover areas of potential discriminatory problems.

For a large company in the home mortgage industry, our experts evaluated the ability of the company’s automated underwriting system (AUS) to predict default, foreclosure, and other adverse outcomes for all groups. We analyzed whether the AUS predictions were equally valid for all groups and whether the underlying statistical model over–predicted minority defaults. We also considered whether there were alternative AUS systems that would have less disparate impact and that could predict default and other loss measures equally well.

On behalf of a national insurance company, our experts served as consulting experts in a discrimination class action matter in which minorities alleged that the company unfairly marketed and issued home insurance policies. Analyzed zip code level variation in market shares and related that variation to factors such as income, house values, home ownership rates, and the minority share of zip codes. Evaluated the extent to which zip codes varied in type of policies underwritten, while distinguishing between high cost full-replacement policies and alternative lower-cost policies sold by the same insurers. In addition to this analysis, our experts worked to rebut claims of opposing experts.