Does anyone know of a company that has successfully defended a Disparate Impact case against the Department of Justice? I can’t think of any. There are a lot of “settlements” but no wins. Why is that?
Well let’s take a look at some of the facts:
- Even the largest financial institutions in the country, with a battalion of attorneys and money to spend on technology, pay tens of millions in settlements.
- The data used to substantiate these claims is generally wrong, yet the data is spun in a way that makes you look guilty whether you are or not.
- The quantity of data fields reviewed does not equal how a loan is priced.
- Regression analysis is flawed because the data is flawed.
- If you are going to subject lenders to fair lending and disparate impact claims then the data used should be spot on correct!
So, if we know that the data is insufficient and does not equal the data required for a rate/price, then one can see that this is a losing battle without the right tools.
I’m not a mathematician, but if I am using insufficient data to formulate a conclusion, my results will be insufficient as well!
Now that we know the problem, we need to formulate the solution, to stop bleeding from our bank accounts over these claims.
Yesterday I talked about simple file notations, that when applied, present the reasoning behind a product selection. Today I am taking a deeper dive into the beginning of the data collection.
Take a look at the options used by many lenders for a compliance management system for pricing, fair lending and disparate impact. You will find that they are spending a lot of money on systems that provide them with insufficient data to defend themselves, because they calculate results based on insufficient data. Not only that, but they do these calculations based on closed loans. Let me ask you this. Do you underwrite a loan after you closed it? That sounds insane doesn’t it? Yet when you review files for fair lending after they close, that is what you are doing.
Effective Compliance Management for Fair Lending Involves Your LOS and Pricing Engine as Data from each is Essential for Correct Results! You have heard the phrase, “It takes a village”. Well in data collection for a fair lending exam it takes the entire village of your front end systems (or at least it does now).
The first step is to take a look at how data enters the loan process. Each time you see “Stop” (below) this means thought needs to be given as to what data fields need collected, and what system this data should go in. I will write about analysis, audit trails and logistics later this week. Right now I am focusing on field definition for each step of the process.
•Client sees an ad, is solicited by phone, or gets referred. STOP
•Client speaks with your retail front line or TPO. This could be the originator directly or an intermediary collecting basic information. STOP
•Client requests a rate, scenario or requests a loan application. STOP
•Client presents enough documents to call it an application. STOP
•Loan is structured, and loan options are reviewed. STOP
•Loan gets locked or registered. STOP
•Loan gets re-locked (potentially numerous times). STOP
•Loan is processed. STOP
•Title Received. STOP
•Appraisal Received. STOP
•Loan is underwritten. STOP
•Conditions are applied and gathered. STOP
•Final compliance review. STOP
•Closing. STOP-At this point the file is closed so it is what it is. However, a final post closing check should be done and action taken if there is a rate/fees violation. Action is restitution to the client for overpayment; according to examiners.
