This report assessed the fair lending practices of “First Community Bank.” A redlining risk analysis compared the Bank's lending distribution to aggregate data, finding no statistically significant differences in minority geographies. Regression analyses of consumer lending data (HMDA data was insufficient), examining pricing and underwriting, revealed no statistically significant disparities based on race, ethnicity, or gender. The analyses used econometric modeling and proxy methods (BISG) to account for missing data and identify protected groups. The conclusion was that the Bank lending practices show no evidence of discrimination.
The information in this case study is based on the provided report and is intended to illustrate Premier Insights, Inc.'s analytical capabilities. The names of the financial institution and market areas have been changed for confidentiality purposes.
Introduction
Premier Insights, Inc. provides comprehensive analytical services to financial institutions, focusing on compliance and fair lending management. This case study details a redlining risk analysis and regression analysis of lending operations conducted for a fictional bank, First Community Bank (“the Bank”) in the market of Rivertown, USA (“Rivertown”). The analysis demonstrates Premier Insights’ expertise in utilizing statistical methods to evaluate lending practices and identify potential areas of concern.
Scope of Analysis
The engagement for the Bank included an examination of lending data within the Rivertown market. The analysis encompassed several components:
- Redlining risk analysis of HMDA data
- Regression analysis of consumer underwriting
- Regression analysis of consumer pricing
Redlining Risk Analysis
- Objective: To assess the geographic distribution of HMDA lending activity, and determine if there were any significant differences in lending penetration for the Bank’s HMDA compared to aggregate HMDA within minority areas.
- Methodology: A statistical approach compared the Bank’s HMDA data by census tract to aggregate HMDA data across the assessment area. This was done at both the assessment area and individual tract levels.
- At the assessment area level, the proportion of the Bank’s lending activity in minority tracts was compared to the same proportions of aggregate HMDA, with statistical tests applied to determine if differences were significant.
- At the tract level, an expected range of applications was estimated for each tract, assuming the Bank's penetration was statistically equivalent to the aggregate. The number of Bank applications was then compared to the expected ranges to identify any underrepresentation.
- Findings: The analysis found that the Bank’s distribution of lending with regard to minority and non-minority geographies was equivalent to that of aggregate HMDA at both the assessment area and tract level. This indicated no statistically significant differences in lending penetration within minority areas. The Bank reported 50 total applications, 39 of which were within the Rivertown area. All of these applications were for non-owner-occupied units.
Regression Analysis of Consumer Lending
Objective: To analyze consumer lending data for underwriting and pricing disparities based on race, ethnicity, and gender. Due to an insufficient number of HMDA records, this analysis was limited to consumer lending data.
- Methodology:
- Proxy Methods: The analysis employed the Bayesian Improved Surname Geocoding (BISG) method, a method recommended by the CFPB, to assign race and ethnicity based on customers’ last names and the racial/ethnic makeup of their area. Gender assignments were based on first names using Census data.
- Econometric Modeling: Multivariate regression models were used to analyze the data, controlling for multiple factors that influence pricing and underwriting. This methodology allowed for the assessment of the independent influence of each variable. Models were constructed based on the Bank’s pricing and underwriting guidelines. Explanatory factors included credit score, DTI, and LTV.
- Target Groups: The analysis focused on three primary target groups: Black applicants, Hispanic applicants, and female applicants.
- Findings:
- Underwriting: The regression analysis of underwriting found no statistically significant disparities in denial rates between target and control groups. The analysis showed that the Bank's underwriting decisions were explained by relevant criteria.
- Pricing: The regression analysis of pricing also found no statistically significant disparities based on race, ethnicity or gender. The analysis indicated that the Bank's pricing decisions were explained by relevant factors.
Conclusion
The analysis conducted for First Community Bank provided valuable insights into its lending practices.
- The Bank used relevant and appropriate criteria for both pricing and underwriting. The findings confirm that, on average, the bank followed its pricing and underwriting guidance as dictated in policy.
- The statistical models and tests did not reveal any discriminatory preferences based on race, ethnicity, or gender in either pricing or underwriting.
- The consistency of estimates further supported the neutrality of prohibited factors in the bank’s decision-making.
This case study demonstrates Premier Insights, Inc.'s capability to conduct thorough analyses of lending data to assess fair lending compliance. Our expertise in statistical methodologies and our commitment to rigorous analysis enable us to provide financial institutions with the insights they need to operate fairly and effectively.