This report assesses the fair lending implications of raising interest rates on River Valley Bank's Community Mortgage (CML) loan program, which serves low-to-moderate income borrowers. The analysis uses one year of loan data and regression models to examine three rate-change scenarios, comparing outcomes for Black and White applicants. Simulations are employed to project the impact of rate increases on the probability of finding statistically significant racial disparities. The findings suggest that smaller rate increases pose minimal fair lending risk if the CML program is properly controlled for in the analysis and consistent pricing practices are maintained. Larger increases, however, significantly elevate that risk.
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
This document presents a case study of a fair lending analysis conducted by Premier Insights, Inc. for a financial institution. The analysis focuses on a proposed rate change to a special loan program and its potential impact on protected classes. To maintain confidentiality, all bank and market names have been fictionalized. The names ONLY have been altered, with all the underlying data, analysis, and strategic approach based on an actual client engagement with Premier Insights. This case study demonstrates Premier Insights' expertise in evaluating fair lending risks and providing data-driven insights for decision-making.
Background
A fictional financial institution, River Valley Bank, developed a special loan program called the Community Mortgage (CML) program, designed to expand credit access for low-to-moderate income (LMI) buyers in the three-city area. This loan program has more flexible underwriting standards than the Bank’s typical loan offerings. The Bank was considering raising the rates for these loans to match the standard rates for similar products and sought to assess potential fair lending implications.
Data and Methodology
- Data: The analysis utilized a calendar year loan data for 1-4 family, owner-occupied, first-lien loans, focusing on customers identified as Black or non-Hispanic White. The sample included 1,074 loans, with 224 to Black applicants and 850 to White applicants.
- Scenarios: Three rate change scenarios were analyzed:
- Scenario 1: Upward adjustments to CML loans, with other loans priced as they were.
- Scenario 2: Assigning rates based on the rate sheet that would have applied on any given day, including raising the rates on CML loans to the standard rate.
- Scenario 3: Increasing the CML loans to the standard rate while leaving all other loans priced as they were.
- Regression Analysis: Regression analysis was used to measure the impact of the rate changes on pricing disparities between Black and White applicants. The target-control group was Black versus non-Hispanic White. The dependent variable was the contract note rate charged for each loan. The analysis included control variables such as loan term and whether the loan was an adjustable-rate mortgage (ARM). Regression models were run both with and without an indicator for the CML program.
- Simulations: In addition to regressions of the actual data, these scenarios were tested in simulation, generating 1,000 random samples for each pricing scenario to understand the impact of changes in the samples over time.
Key Findings
- Initial Disparity: In the initial data, Black applicants were charged an average rate of 3.917%, compared to 3.968% for White applicants. This 0.061% difference was largely because 54% of Black applicants were in the CML program, which had a lower average rate of 3.778%, compared to 14% of White applicants. The average rate for non-CML customers was 4.011%.
- Impact of CML: When controlling for the CML program, the rate difference between Black and White applicants was not significant. The results showed that the CML program provided a rate discount of 0.176%.
- Rate Increase Impact: Any increase to the CML rate would disproportionately impact Black customers since they are more likely to be in the program. A 0.10 increase in the CML rate would increase the black coefficient by approximately 0.04.
- Simulations: Simulations showed that increasing the CML rate by 0.15 when there exists only a 0.04 difference between pricing of black and white applicants can lead to a finding of a statistically significant disparity 70% of the time.
- Test Rate 1: When all loans based on rate sheet terms were assigned rates that would have applied and CML loans were changed to the standard rate, a pricing disparity with Black applicants being charged higher rates emerged.
- Importance of Controls: Having no significant pricing differences outside of the CML product and controlling for the CML product in the models are necessary to avoid fair lending risk from a CML rate increase.
Analysis of Rate Change Scenarios
- Incremental Increases: Incremental increases to the CML rate (0.10, 0.25, and 0.50) showed an increase in the black coefficient, thus increasing the chances of finding a disparity based on race. A 0.25 or greater increase is very likely to result in a significant Black coefficient.
- Test Rates: Test Rate 1 and Test Rate 2, which involved assigning rates based on the rate sheet and raising the rates on CML loans, showed an increase in the black coefficient of approximately 0.30 to 0.40.
Discussion
- The analysis revealed that the existing lower rates for Black borrowers were almost entirely attributable to the discounted pricing of the CML product.
- If the CML loans are accounted for in the data, pricing differences between the two groups are essentially zero.
- Risk: There exists some fair lending risk for the bank in increasing the CML rate. This risk depends largely on whether the CML product is controlled for in the models and the consistency of the bank’s pricing practices. Increasing the CML rate by 0.15 when there exists only a 0.04 difference in pricing leads to a statistically significant disparity 70% of the time.
- Even if strict pricing discipline was enforced, there is still fair lending risk associated with increasing the CML rate unless the increase is 0.10 or less.
- The low R-squared values suggest that other omitted factors such as fluctuations in rates due to time periods and origination fee related discounts may also explain the rates charged.
Conclusion
The analysis conducted by Premier Insights highlighted the importance of considering the potential fair lending implications when making changes to loan programs. The case study demonstrates that while the CML program was intended to serve LMI buyers, changes to its pricing structure could lead to unintended disparities based on race. The analysis suggests that the Bank proceed cautiously and only increase the rates on these loans by 0.10 or less.
This case study demonstrates Premier Insights' ability to provide rigorous, data-driven analysis to help financial institutions mitigate fair lending risk and make informed decisions. The analysis considered multiple scenarios and provided risk assessments based on simulations. The information is presented in a way that can help institutions better understand the potential financial implications of such changes.