WE’RE HEARING quite a few regional banks have been in touch with us lately. They want to ramp up their agency production as they begin to gain mortgage volume share. It makes good sense. These regional banks are likely to be in good shape in our future industry because they have existing clients to sell mortgages to and they have the ability to fund mortgages and hold them in portfolio. But they also want the outlet and ability to sell the loans into the secondary market, and the GSEs are a key part of that.
Of course, the agencies appear more than willing to buy them from these banks, especially since bank owned mortgage entities typically have strong balance sheets and are thus good counterparties for the agencies.
However, you may have read a recent
The article made clear that the agencies are really looking for 100% quality (zero defects) from those that sell them loans. Perfection is a hard thing to expect, especially related to capturing all the required documents and information necessary for a loan. So, here are a few quick observations I have related to the pursuit of perfection.
First off, the agencies consistently tell us that the biggest issue from the loan file is when data is inconsistent, or when the data do not match the documents. By the way, this is one of the same observations made by the CFPB when they do an audit. So, no matter who does the reviewing, the review is often about data integrity. When the data is inconsistent you automatically have a potential “quality” issue when the definition of quality is that you can’t have a defect. The problem with our industry is that the process of capturing the data is often handled by humans reading forms, and humans make mistakes.
So, while the GSEs may expect perfect, that is a hard thing to accomplish when imperfect humans do the work. After all, you are only as strong as your weakest link and some of our staff are likely to be “weak” at reviewing tax returns, or entering 1003 information correctly. The best solution to this may be to eliminate as much of the data entry as possible, not just expect the data to always be entered correctly.
Let me give you an example of how this is problematic. Currently, nearly all credit reports are obtained by the lender automatically, with electronic data provided back to the LOS. Thus, you typically do not have to re-enter the credit information on the 1003; it’s automatically populated. And when that same info is then shared with the automated underwriting engine, the data integrity for credit items is maintained and you likely have consistent data throughout the process. Nearly perfect right?
Imagine for a second you were back in 1980 and the ordering of a credit report was a process that culminated in a written report spitting out of a printer after a human keyed in some information into a small terminal. (It actually printed in triplicate, too, on that neat carbonless paper!) Yes, I know I am dating myself, but that is how we used to get credit reports.
Once the report was received, the processor would manually enter all of the information on it into the LOS machine, and we would start building a loan file. Sounds crazy, right? Well, that is exactly how to we still normally handle income, assets and appraisal. We ask for the documents, we wait until they come back (and hopefully not in triplicate!), we review them, we base our calculations on them and then we enter the data into the system.
Even if you have all electronic images, you still have an imperfect human entering the data from the image into the LOS. The biggest advancement in this process in the past few years has actually been hardware to make the processor work easier (ooh, split screen).
At Stratmor, we have seen some technologies recently that really change that paradigm. They generate data instead of images; they make that data nearly perfect and they make that data available almost instantly. Our analysis is that such processes dramatically reduce the cost of producing the loan and improve pull through and reduce turn times. Wow! They also make such a loan file much closer to the perfect ideal that the agencies expect. So in the pursuit of perfection, I recommend we follow the path of the credit bureau information and get the data right up front.
Feel free to drop me a
Garth Graham is a partner with Stratmor Group, and has over 25 years of mortgage experience, from Fortune 500 companies to startups, including management of two of the most successful mortgage e-commerce platforms. He was formerly with Chase Manhattan Mortgage and ABN Amro, where he was a senior executive during the sale of its mortgage group to Citigroup.
