Diagnosing the Fair Lending Health of Your HMDA-LAR

The New Year brings about all kinds of resolutions, whether they are to read more books or to start that cooking class that has been put off. But the most common resolutions typically involve health. Gym membership numbers explode this time of year as people try to create a habit of working out in an attempt to improve their well-being and shed those extra pounds.

This article discusses starting the New Year with a resolution involving health as well, except this resolution involves the fair lending health of your Home Mortgage Disclosure Act Loan Application Register (HMDA-LAR). There are several simple tests you can integrate into your quarterly HMDA process to help gauge the fair lending health of your HMDA-LAR and identify potential issues before they become problematic.

The information contained in HMDA-LARs offers a plethora of valuable insights. While there are many fields in a HMDA-LAR, only six fields are needed for the three tests discussed below: the ethnicity, race, and sex of the applicants, and the action taken type, the application date, and the action taken date of each application. By using these fields, you can analyze the differences between protected classes and control groups as they relate to ethnicity, race, and sex. For purposes of this article, control groups will always be non-Hispanic (when focusing on ethnicity), white (when focusing on race), and male (when focusing on sex). A protected class is any ethnicity, race, or sex that is not in the control group.

Test #1. The first test looks at the denial rate for protected classes as compared to their related control group. This is one of the most common and valuable ways to look at the fair lending health of your HMDA-LAR. One way in which this can be done is to calculate a disparity ratio. The disparity ratio will tell you the rate at which the protected class is denied in relation to its control group. For example, assume that out of 20 Asian applicants (your protected class), 10 were denied. That equates to a 50% rate of denial. Now when looking at white applicants (your control group), you see that out of 40 applicants, 12 were denied, equating to a 30% rate of denial. To calculate the disparity ratio, take the rate of denial for Asian applicants (50%) and divide it by the rate of denial for white applicants (30%). The resulting disparity ratio of 1.67 means that at face value Asian applicants were 1.67 times more likely to be denied than white applicants.

Is this evidence of a fair lending issue? By itself, no. There are valid reasons why Asian applicants may have been denied at a higher rate than white applicants during the underwriting process. However, the disparity ratio allows you to isolate those protected classes with a significant disparity ratio and perform additional research to identify whether there might be a fair lending issue. What constitutes a significant disparity ratio? While there is no legally established number, a generally accepted standard is that any disparity ratio of 1.50 or greater constitutes a significant disparity ratio worthy of additional research.

Test #2. The second test looks at another facet of fair lending—the processing period of applications. The purpose here is to compare a protected class to a control group to determine whether the protected class faces a longer period of time in the processing of their applications from application date to action date. You will want to take the average processing period for the protected class and compare it to the average processing period for the control group. For example, you can look at all applications received from females and calculate the average number of days between the application date and the action date, regardless of the action type. Then you will perform the same calculation for your control group (in this example, that would be male applicants) and compare the two averages. Similar to the denial disparity ratio calculated in the first test, there is no legally established number of days that would be an immediate trigger for further review. However, anything greater than seven days difference between the protected class and the control group is generally considered as warranting further review.

Test #3. The third and final fair lending test considers the rate of withdrawn and incomplete applications for a protected class compared to the rate of withdrawn and incomplete applications for a control group. For example, assume your HMDA-LAR has applications from 50 Hispanic borrowers (your protected class) with 15 of those resulting in withdrawn applications and another 5 ending up as incomplete. That means that 20 of the 50 applications for Hispanic borrowers were either withdrawn or incomplete, equating to a withdrawn/incomplete rate of 40%. Your control group is nonHispanic applicants, who in this example had 100 entries in your HMDA-LAR. Of those 100 entries, 10 were withdrawn and 2 were incomplete, resulting in a withdrawn/incomplete rate of 12%. Again, while no magic number automatically triggers alarm bells, generally if the withdrawn/incomplete rate for the protected class is two or more times greater than the withdrawn/incomplete rate for the control group (as it is in the example), that can be an indicator that further file review may be warranted. And remember, just because additional file review occurs does not inherently mean there is a fair lending problem. There are many legitimate reasons why a protected class might have a higher withdrawn/incomplete rate than the control group.

The three tests discussed above are not an exhaustive list of the types of fair lending tests that can be performed, and they are not meant to be a substitute for a full fair lending review where underwriting and pricing data across all loan product sets, in addition to a review of lending policies/procedures and marketing efforts, are all analyzed. Rather, they are intended to be used as a relatively quick and simple way to benchmark the fair lending health of your HMDA-LAR. When performed on a regular basis, the tests should help to identify statistical anomalies, and the follow-up file review should provide supporting data to explain them. In addition, this is a fantastic way to show your regulator that you are staying on top of your fair lending game. By sticking with your resolution and making these tests a part of your quarterly HMDA process, the fair lending health of your HMDA-LAR will be much less likely to provide any unwelcome surprises.