With all of the data that is now being collected in 2018, many institutions may not quite know where to start in pulling together fair lending analytics. Congratulations—as “thinking” about how to approach “telling your story” under the new expanded HMDA data set is actually a good place to start.
In 2017, and prior years, HMDA “plus” data may have been part of the information that a financial institution gathered, so there may already be some basic analytics in place that can continue with the new 2018 expanded data set—a good place to start! However, regardless of the analytics approach you used in the past, it is important to consider how those fields may have changed, and what additional information may now be available to develop more robust analysis.
Specifically, within the new expanded 2018 data, there are 25 new data points, with 14 of 23 existing data points modified, and some fields bringing in items that were previously optional, leaving only nine which existed previously and which remain the same. That means there is a lot of new stuff! Thirty-seven (37) of these fields are considered “key data fields” for examination purposes, as designated by the FRB, FDIC and OCC (see for example FDIC FIL-51-2017.) Considering the choices available for each point, there are 110 possible reportable fields.
These expanded data points will enhance the ability of financial institutions, examiners and, to some degree, the public, to perform more in-depth fair lending analytics at any time. But this data expansion may create some new concerns for institutions. For example, examiners will have the ability to pre-determine their focal points—without having worked with the institution—and thus without having the necessary context of the institution’s market, products and services. This means that financial institutions should have a very good understanding of their own analytics, in context, to anticipate any concerns raised by examiners, and be able to readily address any apparent inconsistencies.
At this juncture, we are still awaiting a final determination by the Consumer Financial Protection Bureau (CFPB) regarding its proposal to release public data. As proposed, the way in which HMDA data is made available to the public may create a different set of focal points for institutions to monitor. While most of the fields are proposed to be published as reported, a handful of fields would be reported in a modified format. For example, Age and Debt-to-Income ratios would be reported in a range, and Property Value and Loan Amount fields would also be reported in a modified format.
To put a finer point on the availability of data, there are quite a few new possible combinations of the data available to perform a variety of fair lending analytics. Demographic information (what we used to refer to as Government Monitoring Information or “GMI”) represents roughly one-third of those combinations, and adds quite a lot of new information to be parsed and analyzed between aggregate and disaggregate categories.
Among the new fields of data being reported, the required reporting of dwelling-secured, open-end lines of credit (commonly known as a Home Equity Lines of Credit or HELOCS) may be the single most impactful change of all the new reporting requirements. Inclusion of HELOC data adds an entire product set of data that many institutions may have not analyzed previously. This influx of data may yield some surprising risks for institutions. These risks could be associated with the traditional pricing and underwriting risks as well as redlining risks, as these new application records may skew an institution’s lending pattern. Pre-approval requests—approved but not accepted were previously optional to report, however these records are now required and will add more records to the reporting for those institutions that participate in a pre-approval program. Additionally, specific to the State of New York, applications in which the transaction was completed involving a CEMA (Consolidation, Extension, and Modification Agreement) are now reportable. For New York lenders, this could be a significant portion of the application population, so again, institutions should start early to try to assess the impact of these new fields on their overall data analysis.
While data integrity is a critical consideration across all data fields, it is especially important that the information “relied upon” in making the credit decision be documented for many fields in these areas. These fields include the Loan-to-Value (LTV), Debt-to-Income (DTI), Credit Score and the Automated Underwriting System (AUS) Result. These will all be closely monitored in association with the pricing and underwriting outcomes and institutions should look deeply for unexpected data within these fields.
Another new aspect to fair lending analytics relates to the availability of peer data. While peer data has historically been released in the fall, the Bureau has already released a 2017 peer data set in May. Institutions should take advantage of this early access, perform analytics to understand performance in comparison to their peers, and make adjustments accordingly.
Overall, the 2018 extended data set should be analyzed in as many combinations as is feasible to help institutions obtain an overall picture of its lending performance. Data relating to demographic information should be analyzed against underwriting and pricing criteria to try to determine whether there is any disparate impact or disparate treatment on a prohibited basis.
Regression analysis may help to identify any apparent discrepancies, so institutions can further refine the information available to “tell your story”. Institutions could also be conducting comparative file reviews on outliers and ensure that there is sufficient documentation to support their reasonable business justification for any actions taken.
Institutions should consider all of these changes in order to get a good start in understanding the fair lending impact of the 2018 expanded HMDA data fields. And with this initial approach, institutions are well on their way to develop additional analytics to further refine their “story” as it continues to unfold.