With the announcement, earlier this year about the latest data additions to be included in all future HMDA reporting, the industry has been heavily focused on making sure that the necessary data is available within loan origination systems. Furthermore, the loan application form, commonly referred to as the 1003, has been updated to ensure that all this data can be collected from the borrower(s). This additional information, in conjunction with what is already collected, form the basis of regulators Fair Lending reviews.
Fair Lending, is the federal regulation that requires all lenders to treat every applicant equally. For depository institutions, their lending patters must demonstrate that they offer mortgage opportunities in the communities in which they accept deposits. Additional analysis is also conducted on the areas in which a lender typically lends. This has traditionally been known as the lender’s footprint and is measured by racial population distributions within specific metropolitan statistical areas or MSAs. In other words, if an MSA is 50% Hispanic, regulators would expect to see that 50% of your applicants are Hispanic. This they believe demonstrates the “fairness” of your lending practices. There are however some very “unfair” issues associated with this analysis, many of which will more than likely be exacerbated by the collection of additional data and the scrutiny of the CFPB.
The most obvious of these is the poor quality of the data. Although the submission process includes quality and validity checks, inaccurate and/or inconsistent data is rampant. While most lenders work diligently to ensure good data, there have been instances where manufactured and calculated data have been used. Furthermore, until this past week’s announcement, there has never been a way to identify if all required lenders have even submitted their data. If data is submitted late or corrected and resubmitted, the changes never make it into the overall HMDA database for the year. Imagine one lender’s surprise upon finding out that the entire LAR they submitted one year was not included at all.
Unfortunately, even the bad and or missing data included in the HMDA database is used to analyze lenders. For example, not all applications have the monitoring data completed and since it is the borrowers’ prerogative to complete, few, if any lenders have all the race gender and ethnicity data for every application. This can lead to some very unfair conclusions. Recent comparisons of the number of loan applications compared to these completion of monitoring data found that these numbers just don’t add up. For example, if a lender has 10,000 applications but the breakdown by race shows that only 37% were minority, does that mean that 48% are white? If so, and the population is the MSA is 52% minority does this mean the lender is failing to meet regulatory standards? Without knowing the race of the remaining 15% of the applicants, it is impossible to tell. Yet this is a major part of the regulatory review. Isn’t this a bit deceptive on the part of the regulator?
Finally, regulators and lenders alike must reconsider the use of comparative footprints in conducting this analysis. When lenders and banks were primarily regionalized this may have made sense but with the expansion to nationwide lending and the use of electronic applications, this model is unreliable and in fact deceptive when reaching any conclusion. This must be changed if we are truly to identify any discrimination practices.
The issues identified here are clear indicators that the regulators are not accurately measuring a lender’s Fair Lending, but instead are conducting unfair and deceptive analytics themselves. To protect themselves, it in in every lender’s best interest to know more about their HMDA data then any regulator does.