“It is a capital mistake to theorize before one has data.
Insensibly, one begins to twist facts to suit theories instead
of theories to suit facts.”
The Adventures of Sherlock Holmes, A Scandal in Bohemia
Mortgage lenders finally appear to be seeing the light at the end of the tunnel. After years of negative publicity and what felt like an endless stream of new regulations that resulted in unexpected costs without healthy business opportunities, the picture looks brighter. Nonetheless, most attorneys and industry experts agree that mortgage lenders are still a major target for the Consumer Financial Protection Bureau (CFPB) – and despite the threats from the CFPB that examinations of TILA-RESPA Integrated Disclosures rule compliance are going to be rigorous, the next focus is clearly on fair lending laws and the Home Mortgage Disclosure Act (HMDA).
And this focus is not one to be taken lightly.
As most people in the industry are aware, HMDA requires lenders to report certain data to the federal government each year. This data is then used to determine if lenders are meeting the needs of the communities in which they do business.
As per the CFPB, “The HMDA data set is the leading source of information about the mortgage market. Yet, it has not kept pace with the use of new products and features that are now a regular part of the mortgage market. For example, the data does not provide information about certain loan features such as adjustable-rate mortgages for evaluating borrowers’ creditworthiness.”
Because of this missing data, the CFPB has issued new HMDA reporting requirements that are intended to align reported information with current industry data standards. By doing this, the bureau believes that it will ease lenders’ burden in filing the data, as well as improve the information that regulators use to identify lenders that are in violation of Fair Lending Act requirements.
During the Mortgage Bankers Association’s (MBA) recent Annual Convention and Expo, Richard Cordray, director of the CFPB, stated that this data “tells us how lenders are serving the housing needs of their communities. It gives public officials information to help them decide how best to distribute public-sector investments to attract private investment to areas where it is most needed. It reveals lending patterns that could be discriminatory. And it helps lenders understand changing loan patterns in their own local markets. In general, this rule provides better information on the mortgage market, including the reporting of data elements that will help regulators, industry and the public to understand the effects of the recent reforms.”
He went on to say that the CFPB has found that redlining (the inconsistent treatment of applicants based on their protected classes) continues to exist and that the elimination of it is a priority for the bureau’s supervisory efforts. In addition, rather than submitting HMDA data to the Federal Financial Institutions Examination Council (FFIEC), lenders are now required to submit their data directly to the CFPB.
Based on these remarks and others, it is very clear that going forward, HMDA data will be subject to greater scrutiny from regulators. For lenders, it is even more critical to ensure that the data submitted is accurate and, as recommended by numerous speakers at the MBA’s recent Regulatory Compliance Conference, that lenders know more about their data and what it says than the regulators do. After all, with the increased scope of information that is required for future reporting, regulators will have a much more comprehensive picture of a lender’s credit decisions – and, thereby, lending patterns and practices – for conducting a fair lending analysis.
Lenders, however, must learn to be aware of what lies within this data. Those that rely on simple policies and procedures, without evaluating their implementation and adherence to these policies and procedures, are at the greatest risk for examination and unlawful fair lending accusations.
Garbage in, garbage out
Because “mortgage data” is fundamental to HMDA reporting and the foundation for any claims of fair lending failures, one must have an understanding of exactly how much data we are talking about. According to Mortgage TrueView, a data analytics firm that specializes in HMDA, the HMDA data set over the past six years alone comprises approximately 93 million records, with approximately 50 fields for each record. Each field, or cell, has values that range from two possible options (e.g., “HOEPA loan” or “not a HOEPA loan”) to more than 70,000 values (e.g., “census tract”), resulting in trillions of possible data dimensions to be evaluated in understanding both the overall market and a lender’s “story.” With this much data, it seems only logical that the data be reported consistently and accurately.
To determine if, in fact, what is currently reported is actually accurate and consistent, Mortgage TrueView evaluated the existing HMDA data files. To accomplish this task, the company requested and obtained copies of the original HMDA data files from numerous lenders. Once these files were obtained, they were compared with the public loan application register (LAR), and inconsistencies and problems were identified. The firm repeated this task for several years, and what it found does not bode well for lenders subject to any type of regulatory review. Even though every lender’s data undergoes a validity and quality test, this review found an overwhelming number of inaccurate, inconsistent and incomplete data in the released HMDA reports.
Among the issues identified was the failure of the FFIEC (to which the data was previously sent) to include updated data. Data that had been corrected or expanded after the filing date was not included in the publicly filed report. For example, there were four respondents that had reported in 2010, 2011 and 2012 but failed to report in 2013. Then, they showed up again in 2014. Also in 2014, there were 10 respondents that filed but had failed to respond in one of the other previous four years. Does this mean that for one year, they stopped doing business? Or does it mean that they did not bother to file for that year? In another instance, one lender that provided the LAR in a timely manner discovered that its data was not even included in the final public HMDA report. This, of course, was not identified by the regulators either, as they acknowledge that they lack the analytics to identify non-filers.
Further analysis of the data resulted in Mortgage TrueView developing a series of algorithms that allow for comparative analyses to garner a more accurate picture of lending patterns. Using such analyses, the firm identified one lender that, when management was questioned, acknowledged that it “forgot” to include 25,000 loans in its filing.
As for the data, itself, there are numerous issues, some of which are associated with the regulators to which the entity reports. For example, the reporting of HMDA denial reason codes (DRCs) is currently optional for institutions that are not subject to Office of the Comptroller of the Currency regulations. Yet, there have been lenders that have voluntarily reported this information. The results of this analysis show a steady decrease in the percentage of loan denial reasons regardless of the regulatory body to which the lender was reporting. In aggregate, between 2010 and 2014, submission of DRCs dropped by more than 11% – from 79.89% to 68.00%.
For those lenders reporting DRCs, this analysis shows that the use of collateral-related DRCs as the primary DRCs has declined from 28.97% in 2010 to 21.75% in 2014. Applicant-related DRCs have increased from 56.63% to 68.66%, due primarily to an increase in use of the “credit history” DRC (from 25.24% to 35.58%) and the “debt-to-income” DRC (from 25.45% to 27.70%).
What does this mean to any individual lender? Does it mean that fair housing requirements are being violated? Does this result reflect new credit standards, as required by the CFPB? Of more concern is the question of whether results that are inconsistent with this trend precede a CFPB-HMDA and/or fair lending examination. Should any lender whose results are different from the aggregate be concerned?
What’s a lender to do?
With trillions of opportunities to find data that is indicative of regulatory violations, what should a lender do? Just getting all of the data that is now required seems to be the present focus of those involved with this issue. Individuals involved in the data collection from a technology viewpoint do not seem too concerned, as they point to the fact that this is just more data to add to an existing data set. Attorneys and other industry support folks, however, are warning that if lenders are not on top of their data, they could face multiple exams, fines and penalties. In addition, the reputational risk associated with having been found a lender with unfair lending practices can be very severe. Lenders associated with these labels are most likely to find that even the secondary market is concerned about purchasing the loans, let alone the negative reflection from all types of social media.
Of course, the best action for all lenders is to ensure that their data and what it says about them is accurate and reflects their focus on fair lending activities. But with the extensive amount of data that must be reported and that will be used by regulators, where does a lender start? Today, there are a variety of companies that claim to aid lenders in this endeavor. Unfortunately, many times, the information provided is too complex or lacks a format that allows one to make sure one is getting what one needs. Way too often, companies that claim to provide analytical information simply load the HMDA data into a generic business intelligence platform and put together some relatively simplistic charts and graphs.
One of the most obvious examples is the analysis that continues to be used repeatedly by many of these companies. It is a peer-to-peer comparison based on specific metropolitan statistical areas (MSAs). Developed when reporting was limited to banks and other regulated entities that were, at most, lending in specific regions, the MSA analysis was helpful. However, with the expansion of reporting to include independent lenders, the growth of national-based lending and the introduction of electronic applications, this type of analysis does very little, if anything, to assist lenders in analyzing their fair lending practices. Yet, this continues to be the most frequently demonstrated type of analysis conducted by many of these companies.
Because many of the results coming from the analysis of the data are based on race, the area of gender discrimination is frequently overlooked. With the growing number of single female applicants, many who are single mothers, the existing data tells us the gap between them and the typical white, male applicant is significant and may, in fact, be greater than that between minority male applicants and their white counterparts. In fact, at a recent regulatory compliance conference, a representative of the CFPB acknowledged that this issue was something the bureau is evaluating. Unfortunately, the companies offering this type of analysis are few and far between.
There are numerous other examples in which the lack of in-depth analysis creates opportunities for regulators to scour the data and find weaknesses in any lender’s operations. Unfortunately, the number of firms providing so-called HMDA analysis weakens the industry and fails lenders by distracting them from substantive insights that enable them to go “toe to toe” with their regulators in a constructive and productive fashion. To do this, lenders must extract the data necessary using complex algorithms that simplify the complex nature of all of this data.
There is no doubt that the industry needs leadership in this area. Although the effort to ensure the data required for reporting continues to be the focus of the industry in general, the more critical factor, the interpretation of what this data is telling the regulators, seems to have been forgotten. Lenders are looking for this type of leadership in addressing a range of HMDA-related issues, but finding any that are doing something about it is difficult.
Without a doubt, the expansion of HMDA data collection and its use by industry regulators presents a hidden danger to lenders. To address this risk, lenders need to begin acting by finding ways to analyze their data to find out what lies within. Previous practices of evaluating HMDA data to avoid the risk of fair lending concerns will no longer suffice. Just scratching the surface of the data and drawing conclusions based on this brief review could very well result in problems never envisioned.
It will be the wisest lenders that take heed of Sherlock Holmes’ advice to ensure that they are not twisting data to support conclusions. After all, no lender wants that “light at the end of the tunnel” to turn out to be a regulatory freight train headed in its direction.
Becky Walzak heads her own consulting firm, rjbWalzak, and is president of Looking Glass Group, a consultancy that focuses on strategic planning, risk assessment, and operational and sales/marketing services for mortgage lenders. With more than 30 years of experience in the financial services industry, Walzak is an expert in loan quality assurance and risk management. She can be reached at firstname.lastname@example.org.